TWI536806B - System and method for multi-stream video compression using multiple encoding formats - Google Patents

Description

System and method for multi-stream video compression using multiple encoding formats

The present invention is generally in the field of data processing systems that improve the user's ability to manipulate and access audio and video media.

The present application claims priority to US Provisional Application No. 61/210,888, filed on March 23, 2009, entitled "System & Method For Compressing Video Using Feedback," filed on January 23, 2009 A part of the continuation application of US Application No. 12/359,150, filed in the same application, which is filed on December 5, 2007. The continuation application of US Application No. 11/999,475, entitled "Hosting And Broadcasting Virtual Events Using Streaming Interactive Video", filed on December 10, 2002, entitled "Apparatus and Method for Wireless Video Gaming, application No. 10/315, 460, part of the continuation (CIP) application, which is assigned to the assignee of this CIP application.

Recorded audio and animated media have become a part of society since the days of Thomas Edison. At the beginning of the 20th century, recorded audio media (magnetic columns and recorders) and animated media (nickelodeon and movies) were widely distributed, but the two technologies were still in their infancy. In the late 1920s, animations were combined with audio in large numbers, which in turn produced color animations with audio. Radio broadcasts have evolved into broadcast mass market audio media in the form of advertising support. After the establishment of television (TV) broadcasting standards in the mid-1940s, broadcast television brought previously recorded animations or live animations into the home as a form of broadcast mass market media.

By the mid-20th century, a large percentage of American households had a record player for playing recorded audio media, a radio for receiving live broadcast audio, and a television for playing live broadcast audio/video (A/V) media. These three "media players" (record players, radios, and TVs) are often combined in a cabinet that shares a common speaker, making it the "media center" of the family. Although media choices are limited for consumers, the media "ecosystem" is fairly stable. Most consumers know how to use the Media Player and are able to fully enjoy the capabilities of these media players. At the same time, media publishers (mainly animation and television studios, and music companies) are able to distribute their media to theaters and homes without extensive piracy or "secondary sales" (ie, resale of old media) )problem. Typically, the issuer does not derive revenue from the secondary sale, and therefore, the secondary sale reduces the revenue that the issuer may additionally receive from the purchaser of the newly sold old media. Although there are certainly old records sold during the mid-20th century, such sales have no major impact on record distributors, because a music track may be listened to differently than an animation or video program that adults usually watch once or only a few times. Hundreds or even thousands of times. Therefore, music media is less susceptible to bad than animation/video media (ie, music media has a retention value for adult consumers). Once a record is purchased, if the consumer likes the music, the consumer may retain the music for a long time.

Since the mid-20th century, the media ecosystem has experienced a series of fundamental changes in the interests and losses of consumers and publishers. With the widespread introduction of audio recorders (especially cassettes with high-quality stereo), the convenience of consumers is undoubtedly higher. But this also marks the general practice of consumer media: the beginning of piracy. There is no doubt that many consumers use cassette tapes to record their own recordings for convenience, but more and more consumers (for example, students in a group dormitory who are ready to access each other's album collection) will make pirated copies. copy. In addition, consumers use tapes to record music played on the radio instead of buying albums or tapes from the publisher.

The emergence of consumer VCRs has led to more consumer convenience, as a VCR can now be set up to record TV shows that can be viewed at a later time, and the VCR has also led to the formation of a video rental industry where movies can be accessed as needed and TV show. The rapid development of mass media home media devices since the mid-1980s has led to unprecedented consumer choice and convenience, and has led to a rapid expansion of the media distribution market.

Today, consumers face excess media choices and excess media devices, many of which are tied to specific media formats or specific publishers. A greedy media consumer may have a stack of devices connected to the TV and computer in each room in the home, resulting in a "rat's nest" cable to one or more televisions and/or personal computers (PCs) And a set of remote controls. (In the context of the context of this application, the term "personal computer" or "PC" refers to any type of computer that is suitable for use in our home or office, including desktop computers, Macintosh. Computer or other non-Windows computer, Windows compatible device, Unix variant, laptop, etc. These devices may include video game consoles, VCRs, DVD players, audio surround sound processors/amplifiers, satellite video converters, cable TV converters, and the like. Moreover, for greedy consumers, there may be multiple devices with similar functions due to compatibility issues. For example, consumers may own both HD-DVD players and Blu-ray DVD players, or Microsoft Xbox. Video game system and Sony Playstation Both video game systems. In fact, due to the incompatibility of some games on the version of the game console, consumers may have XBox and newer versions, such as the Xbox 360. . Consumers are often confused about which video input and which remote control to use. Even if you place the disc in the correct player (for example, DVD, HD-DVD, Blu-ray, Xbox, or Playstation), select the video and audio input for the device, and find the correct remote control, then consume Still facing technical challenges. For example, in the case of a wide-screen DVD, the user may have to first determine and then set the correct aspect ratio of their television or monitor screen (eg, 4:3, Full, Zoom, Wide) Screen Zoom (Wide Zoom), Cinema Wide, etc.). Similarly, the user may have to first determine and then set the correct audio surround sound system format (eg, AC-3, Dolby Digital, DTS, etc.). Consumers are often unaware that they may not be able to enjoy media content with the full capabilities of their television or audio system (for example, watching a movie suppressed in the wrong aspect ratio, or listening to audio in stereo rather than in surround sound).

Internet-based media devices have increasingly been added to the device stack. Such as Sonos The audio device of the digital music system directly receives audio from the Internet in a streaming form. Similarly, as the Slingbox ^TM entertainment player record video and receiving device via a home network in the form of streaming video or the output video via the Internet in a streaming form, via Internet distally on a PC Watch the video. Internet Protocol Television (IPTV) services provide cable-like services via digital subscriber line (DSL) or other home Internet connections. Recently, efforts have also been made to integrate multiple media functions into a single device, such as Moxi running the Windows XP Media Center version. Media Center and PC. While each of these devices provides elements that facilitate the functions performed by the device, each device lacks a ubiquitous and simple method of accessing most of the media. In addition, the cost of manufacturing such devices is often as high as hundreds of dollars due to the need for expensive processing and/or area storage. In addition, such modern consumer electronic devices typically consume a significant amount of power (even when idle), which means that such devices are costly and waste energy resources over time. For example, the device can continue to operate if the consumer forgets to turn off the device or switch to a different video input. Moreover, because none of these devices is a complete solution, the device must be stacked with another device in the home, leaving the user with a mouse-like wire and many remote controls.

Moreover, while many newer Internet-based devices are functioning properly, such devices typically provide media in a more general form than would otherwise be possible. For example, devices that receive video over the Internet in streaming form often receive video material only in streaming form, but do not receive interactive "extra content", such as video and games, that are often accompanied by DVDs. Production" or director commentary. This is due to the fact that interactive materials are often produced in a specific format for a particular device and are intended to be used to handle interactivity at the local end. For example, each of DVD, HD-DVD, and Blu-ray discs has its own specific interactive format. Any home media device or local computer that can be developed to support all popular formats requires a degree of sophistication and flexibility that is likely to make the consumer's operations too expensive and too complex.

The problem is that if a new format is introduced in the future, the local device may not have the hardware capability to support the new format, which means that the consumer has to purchase an upgraded local media device. . For example, if a higher resolution video or stereo video is introduced at a later date (for example, one video stream per eye), the local device may not have the computing power to decode the video, or the local device may not It has hardware for outputting video in this new format (for example, it is assumed that stereoscopic vision is transmitted via 120 fps video (60 fps to each eye) synchronized with shuttered glasses, if the consumer's video hardware can only support 60 fps video, this option is not available without purchasing upgraded hardware).

The degradation and complexity of media devices has become a serious problem in the face of sophisticated interactive media, especially video games.

Divide the modern video game application into four main non-portable hardware platforms: Sony PlayStation 1, 2 and 3 (PS1, PS2 and PS3); Microsoft Xbox And Xbox 360 ; and Nintendo Gamecube And Wii ^TM; and based on the PC game. Each of these platforms is different from one another, so games written to be executed on one platform typically cannot be executed on another platform. There may also be compatibility issues between the first generation of devices and the next generation of devices. Even though most software game developers create software games designed independently of a particular platform to perform a specific game on a particular platform, a software-specific layer (often referred to as the "game development engine") is needed to make the game Suitable for use on a specific platform. Each platform is sold to the consumer as a "console" (ie, a separate box attached to a TV or monitor/speaker), or is itself a PC. Typically, such video games are sold in optical media such as Blu-ray DVD, DVD-ROM or CD-ROM, which contain video games embodied as a sophisticated instant software application. As home broadband speeds have increased, video games are increasingly available for download.

Due to the real-time nature of high-definition video games and high computing demands, the specific requirements for achieving platform compatibility with video game software are extremely strict. For example, we may want full game compatibility from a generation of video games to next-generation video games (for example, from XBox to XBox 360, or from Playstation 2 ("PS2") to Playstation 3 ("PS3"), just like Productivity applications (for example, Microsoft Word) have general compatibility between a PC and another PC with a faster processing unit or core. However, this is not the case for video games because video game manufacturers are usually releasing a generation. Video games seek the highest possible performance at a given price point, so significant architectural changes are often made to the system, so many games written for previous generation systems cannot work on later generation systems. For example, XBox is based on x86 series processing. The XBox 360 is based on the PowerPC series.

Techniques can be used to simulate a previous architecture, but if the video game is an instant application, it is often difficult to achieve the exact same behavior in the simulation. This is not good for consumers, video game console manufacturers and video game software publishers. For consumers, it means that it is necessary to keep the old and new video game consoles connected to the TV to play all the games. For the console manufacturer, it means the cost associated with the simulation and slower adoption of the new console. For publishers, it means that multiple versions of new games may have to be released to get all possible consumers, not only for each brand of video games (eg, XBox, Playstation), but also often A version for each version of a given brand (for example, PS2 and PS3). For example, a separate version of Electronic Arts' "Madden NFL 08" was developed for XBox, XBox 360, PS2, PS3, Gamecube, Wii and PC in the platform.

Portable devices such as cellular ("mobile") phones and portable media players also challenge game developers. More and more of these devices are connected to the wireless data network and are capable of downloading video games. However, there are a variety of mobile phones and media devices on the market that have a wide range of display resolutions and computing power. Moreover, because such devices typically have power consumption, cost, and weight constraints, such devices typically lack advanced graphics acceleration hardware such as graphics processing units ("GPUs"), such as those manufactured by NVIDIA of Santa Clara, CA. As a result, game software developers typically develop a given game title for many different types of portable devices. The user may find that a given game title cannot be used for its particular mobile phone or portable media player.

In the case of a home game console, hardware platform manufacturers typically pay royalties to software game developers in order to be able to publish a game on their platform. Mobile phone wireless carriers also typically pay royalties to game publishers in order to download a game to a mobile phone. As far as PC games are concerned, there is no need to pay royalties for the release of games, but game developers often face high costs due to the high user service burden to support a wide range of PC configurations and possible installation issues. In addition, PCs often present less impediment to piracy of game software, as PCs are easily reprogrammed by skilled users, and games may be more susceptible to piracy and easier to distribute (eg, via the Internet). Therefore, for software game developers, there are costs and disadvantages in the release of game consoles, mobile phones and PCs.

For game publishers of consoles and PC software, the cost does not end there. In order to distribute the game through the retail channel, the issuer asks the retailer to pay a wholesale price below the sale price to make it profitable. The issuer also typically has to pay for the physical media that manufactures and distributes the game. Retailers also often require issuers to pay "price protection fees" for possible incidents (such as situations where the game cannot be sold, or where the game is down, or the retailer must refund part or all of the wholesale price and/or purchase The case of recycling the game) pays the fee. In addition, retailers often require publishers to pay fees to help sell games in the form of advertising flyers. In addition, retailers are increasingly repurchasing games from users who have finished playing games, and then selling the games as used games, which typically do not share the old game revenue with the game publisher. Games are often pirated and distributed over the Internet for users to download and make free copies, which adds to the cost of the game publisher.

As the Internet's broadband speeds are increasing and broadband connections are becoming more common in the US and around the world (especially for homes and Internet cafes that lend to Internet-connected PCs), Games are increasingly distributed to PCs or consoles via downloads. In addition, broadband connections are increasingly used to play multiplayer and large-scale multiplayer online games (both of which are referred to in the present invention by the abbreviation "MMOG"). These changes reduce some of the costs and issues associated with retail distribution. Downloading online games solves some of the disadvantages of game publishers because the cost of distribution is usually small and there is almost no cost of unsold media. But the downloaded games are still pirated, and because of their size (often many billions of bytes), downloading these games can take a long time. In addition, multiple games can be filled with small drives, such as those sold with portable computers or with video game consoles. However, the game or MMOG requires an online connection to make the game playable, and the piracy problem is alleviated because the user is usually required to have a valid user account. Unlike a linear media (eg, video and music) that can be copied by a camera that captures the video of the display or a microphone that records audio from the speaker, each video game experience is unique and cannot be used with simple video/audio recordings. copy. Therefore, even in areas where copyright law is not strictly enforced and pirated, the MMOG can be protected from piracy and the business can be supported. For example, Vivendi SA's "World of Warcraft" MMOG has been successfully deployed worldwide without being pirated. As a result, many online or MMOG games (such as Linden Lab's "Second Life" MMOG) generate revenue for gamers via an economic model built into the game where assets are available for trading and can even be generated using online tools. Thus, mechanisms other than the purchase or subscription of conventional game software can be used to pay for the use of online games.

Although online piracy is often reduced due to the nature of online or MMOG, online gamers still face remaining challenges. Many games require substantial local (ie, at home) processing resources to make the online or MMOG work properly. If the user has a low-performance local computer (for example, a computer without a GPU, such as a low-end laptop), it may not be able to play the game. In addition, as the game console gets older, it lags far behind current state of the art and may not be able to handle more advanced games. Even if the user's native PC is able to handle the computational needs of a game, there is often installation complexity. There may be drive incompatibility (eg, if a new game is downloaded, it may install a new version of a graphics driver that renders a previously installed game that is dependent on an older version of the graphics driver inoperable). As more and more games are downloaded, the console can use the local disk space. Over time, complex games are developed from the game, usually after the bug is found and fixed, or the game is modified (for example, if the game developer finds a game level too difficult or too easy to play). Receive the downloaded patch. These patches require a new download. But sometimes, not all users complete the download of all of these patches. At other times, these downloaded patches introduce additional compatibility or disk space consumption issues.

In addition, during game play, a large amount of data download may be required to provide graphical or behavioral information to the local PC or console. For example, if a user enters a room at a MMOG and encounters a scene that is not available on the user's local machine or has a behavior that is not available on the user's local machine or For a role, you must download the information for the scene or role. If the internet connection is not fast enough, this can result in a substantial delay during game play. Moreover, if the scene or character encountered needs to exceed the storage space or computing capacity of the storage space or computing power of the local PC or the console, it may cause the user to continue playing the game or must continue playing the game with the reduced quality graphic. The situation. Therefore, online or MMOG games often limit their storage and/or computational complexity requirements. In addition, such games often limit the amount of data transfer during the game. Online or MMOG games can also narrow the market for users who can play such games.

In addition, the skilled person is increasingly reversing the original copy of the engineering game and modifying the games so that they can cheat. Cheating can be as simple as pressing a button more quickly than humans (for example, to shoot very quickly). In games that support in-game asset trading, cheating can lead to a degree of deception that leads to fraudulent transactions involving assets of real economic value. When an online or MMOG economic model is based on the trading of such assets, this can have substantial negative consequences for the game industry.

As PCs and consoles produce ever-improving games (eg, with more realistic graphics, such as instant ray tracing, and more realistic behaviors, such as real-time simulations), the cost of developing a new game has grown. In the early days of the video game industry, video game development was a process similar to the development of application software; that is, with graphics, audio and behavioral elements or "assets" (such as for an animation with a lot of special effects). Most development costs are spent on software development compared to the development of graphics, audio and behavioral elements or "assets". Today, compared to software development, many of the best video game development efforts are more similar to animation developments with special effects. For example, many video games provide simulations of the 3D world, and increasingly produce photo-realistic characters, props, and environments (ie, computer graphics that appear to be as realistic as real-life images captured by photography). One of the most challenging aspects of the development of a photo-realistic game is the formation of a computer-generated face that is difficult to distinguish from a real-life face. As developed by San Francisco, CA Mova of the Contour ^TM Reality Capture face capture of high-resolution techniques to capture and track the precise geometry of a performer's face is in motion. This technique allows for the reproduction of a 3D face that is substantially indistinguishable from a captured live view on a PC or game console. Accurately capturing and reproducing a "photorealistic" face is useful in several ways. First of all, celebrities or athletes (often hired at high cost) are often used in video games, and the user can be noticeable, making the viewing experience distracting or unpleasant. A high degree of detail is often required to achieve a high degree of photorealisation - many polygons and high resolution textures need to be reproduced, and as the face moves, the polygons and/or textures may change from frame to frame.

When the scene with high polygon counts of detailed texture changes rapidly, the PC or game console supporting the game may not have enough RAM to store enough polygons and texture data for the required number of animated frames generated in the game segment. . In addition, a single optical drive or single drive that is typically available for PCs or game consoles is typically much slower than RAM and typically does not keep up with the maximum data rate that the GPU can accept when rendering polygons and textures. Current games typically load most of these polygons and textures into RAM, which means that the complexity and duration of a given scene is largely limited by the capacity of the RAM. For example, in the case of facial animation, this can limit the PC or game console to a non-photorealistic low-resolution face, or to a photo-realistic face, which can only pause and load more images in the game. The polygons and textures (and other materials) of the box were previously animated for a limited number of frames.

When the PC or console displays a message similar to "Loading", watching a progress bar slowly move through the screen as an inherent shortcoming and is accepted by users of today's complex video games. Self-disk (unless otherwise defined, the term "disk" in this document refers to the delay in loading the next scene to non-volatile optical or magnetic media, as well as non-magnetic media such as semiconductor "flash" memory). Can be a few seconds or even a few minutes. This delay is a waste of time and can be quite frustrating for gamers. As discussed previously, most or all of the delay can be attributed to the loading time of the polygon, texture, or other material from the disk, but it can also be part of the load time for the processor in the PC or console and/or The case where the GPU prepares data for the scene. For example, a soccer video game may allow players to choose among many players, teams, sports fields, and weather conditions. Thus, depending on the particular combination selected, the scene may require different polygons, textures, and other materials (collectively referred to as "objects") (eg, uniforms of different teams have different colors and patterns). It is possible to enumerate as many or all of the various arrangements and pre-calculate as many or all of the objects in advance, and store the objects on a disk for storing the game. However, if the number of permutations is large, the amount of storage space required for all of the objects may be too large to fit on the disk (or too impractical to download). As a result, existing PC and console systems are often constrained in both the complexity of the given scenario and the duration of the game and suffer from the long load times of complex scenes.

Another significant limitation of prior art video game systems and application software systems is that such systems are increasingly using, for example, 3D objects (such as polygons and textures) that must be loaded into a PC or game console for processing. ) a large database. As discussed above, such databases can take a long time to load when stored on a disk at the local end. However, if the database is stored in a remote location and accessed via the Internet, the load time is typically longer. In this case, downloading a large repository can take minutes, hours, or even days. In addition, such databases are often generated with great overhead (for example, a detailed 3D model of a high mast-equipped sailboat used in games, movies, or historical documentaries) and are intended to be sold to local end users. However, the data is inventory at risk: it is pirated once it is downloaded to the local user. In many cases, a user only wants to download a database by evaluating a database to see if it is appropriate for the user (eg, when the user performs a particular move, 3D of a game character) The garment has a satisfactory appearance or appearance.) The long load time can be an obstacle for the user to evaluate the 3D database before deciding to purchase.

Similar problems arise in MMOG, especially allowing users to increasingly use custom character games. If a PC or game console wants to display a character, it must access the 3D geometry (polygon, texture, etc.) and the behavior of the character (for example, if the character has a shield, regardless of whether the shield is strong enough A database to deflect a spear). Usually, when a user plays an MMOG for the first time, many of the character's databases are available in the initial copy of the game, which can be obtained or downloaded to a disk on the game's disc. However, as the game progresses, if the user encounters a character or object that the database cannot obtain at the local end (for example, if another user has generated a customized character), the user or the object must be downloaded before the character or object can be displayed. Its database. This download can cause a substantial delay in the game.

Another challenge for video game developers and publishers of prior art video game consoles is that it takes two to three years and tens of millions of dollars to develop a video game, given the sophistication and complexity of video games. . If the new video game console platform is introduced at a rate of approximately once every five years, the game developer must start the development of the game a few years before the new game console is released, so that the video game is still released on the new platform. Available. Several consoles from competing manufacturers are sometimes released at about the same time (for example, within a year or two), but the popularity of each console is left to be viewed, for example, which console will generate the highest video game software sales. . For example, in a recent console loop, Microsoft XBox 360, Sony Playstation 3, and Nintendo Wii are scheduled to be introduced in approximately the same time frame. But in the years before such introductions, game developers essentially had to be "betting" on which console platforms would be successful than other platforms, and investing their development resources accordingly. The animation production company must also allocate its limited production resources based on its appropriate estimate of the likelihood of success of the film prior to its release. If the level of investment required for video games is growing, game production becomes more and more like animation, and game production companies routinely invest in their production resources based on their estimates of the future success of a particular video game. However, unlike an animation company, this bet is not based solely on the success of the production itself; the truth is that it is based on the success of the game console intended to execute the game. The simultaneous release of multiple console games reduces risk, but this extra effort adds expense and often delays the actual release of the game.

The application software and user environment on the PC has become more computationally intensive, dynamic and interactive, which not only makes it more visually appealing to the user, but also makes it more useful and intuitive. For example, the new Windows ^VistaTM operating system and Macintosh Subsequent versions of the operating system are visually animated. Such as those from Autodesk, Inc. Of Maya ^TM advanced graphics tools provide a very sound of 3D rendering and animation capabilities, these tools increase the limitations of current technology level of the CPU and GPU. However, the computational demands of these new tools create many practical problems for users and software developers of these products.

Since the visual display of an operating system (OS) must work for a wide range of computers (including previous generation computers that are no longer sold but still available for new OS upgrades), the OS graphics needs are largely targeted by the OS. The least commonality of computers, these computers usually include computers that do not include a GPU. This strictly limits the graphics processing power of the OS. In addition, battery-powered portable computers (eg, laptops) limit the visual display capability due to the high computing activity in the CPU or GPU that typically results in higher power consumption and shorter battery life. Portable computers typically include software that automatically reduces processor activity to reduce power consumption when the processor is not being utilized. In some computer models, the user can manually reduce processor activity. For example, Sony's VGN-SZ280P laptop has a switch labeled "Stamina" on one side (for low performance, long battery life) and a switch labeled "Speed" on the other side (applicable) For high performance, short battery life). Even when the computer is executed with a small portion of its peak performance capabilities, an OS executing on a portable computer must be able to operate effectively. As a result, OS graphics processing performance often remains much lower than the computing power available at the current state of the art.

High-end, compute-intensive applications such as Maya are often sold to expect them to be used on high-performance PCs. This typically results in a much higher performance and the least common point requirement for more expensive and less portable. As a result, these applications have more limited target audiences than general-purpose OSs (or general-purpose productivity applications such as Microsoft Office), and their sales are typically lower than general-purpose OS software or general-purpose application software. Potential audiences are further limited because it is often difficult for prospective users to try out these computationally intensive applications in advance. For example, suppose a student wants to learn how to use Maya or a potential purchaser who already knows about such applications wants to try Maya before investing in the purchase (which is likely also involves buying a high-end computer capable of executing Maya). Although the student or the potential purchaser can download a demo version of Maya or obtain a physical media copy of one of the demo versions, if the student or the potential purchaser does not have a computer capable of fully performing Maya (eg, handling a complex 3D scene) The student or the potential purchaser will not be able to obtain an assessment based on all reliable information about the product. This essentially limits the audience for these high-end applications. It also contributes to a high sales price because development costs are usually amortized over much less purchases than general-purpose applications.

High-priced apps have also caused more incentives for individuals and businesses to use pirated copies of their application software. As a result, high-end application software has been pirated, although publishers of the software have tried to reduce this piracy through various technologies. Still, even when using pirated high-end applications, users can't avoid buying expensive current state of the art PCs to perform pirated copies. Therefore, although the user can obtain the use of the software application in a small portion of the actual retail price of the software application, the user of the pirated software is required to purchase or obtain an expensive PC to make full use of the application.

The same is true for users of high-performance pirated video games. Although pirates can obtain such games in a fraction of the actual price of such games, they still need to purchase the expensive computing hardware necessary for normal game play (eg, GPU-enhanced PCs, or high-end video games such as XBox 360). Console). If the video game is usually a pastime for the consumer, the extra cost of the high-end video game system can be inhibited. This situation is worse in countries where the average annual income of workers is very low relative to the average annual income of American workers (for example, China). As a result, a small percentage of people have high-end video game systems or high-end PCs. In these countries, "Internet cafes" (where users can pay to use computers connected to the Internet) are quite common. Such internet cafes often have older models or low-end PCs that do not have high performance features such as GPUs, which can additionally enable players to play computationally intensive video games. This is a key factor in the success of games played on low-end PCs, such as Vivendi's "World of Warcraft," which is very successful in China and is commonly played in Internet cafes. In contrast, it is unlikely to play compute-intensive games like "Second Life" on PCs installed in Internet cafes in China. Users who only have access to low-performance PCs in Internet cafes cannot access these games.

There are also obstacles to users who are considering purchasing a video game and are willing to try out a demo version of the game by downloading the demo to their home computer via the Internet. A video game demonstration is often a fully mature version of the game, but some of the features are disabled or limited in the amount of game play. This may involve a long process (possibly several hours) of downloading a few gigabytes of data before it can be installed on a PC or a console and executed. In the case of a PC, the process may also involve knowing which special drivers (eg, DirectX or OpenGL drivers) the game requires, downloading the correct version, installing the driver, and then determining if the PC can play the game. This latter step may involve determining if the PC has sufficient processing (CPU and GPU) capabilities, sufficient RAM, and a compatible OS (eg, some games are executed on Windows XP, but not on Vista). Therefore, after a long process of attempting to perform a video game demonstration, the user is likely to find that the video game demonstration may not be played due to the user's PC configuration. Worse, once the user has downloaded the new driver to try out the demo, these driver versions may not be compatible with other games or applications that the user typically uses on the PC, so a demonstration installation may make the previous Operational games or applications are not operational. These obstacles not only hinder users, but also become obstacles for video game software publishers and video game developers to sell their games.

Another problem that leads to economic inefficiency is related to the fact that a given PC or game console is typically designed to accommodate the specific performance requirements of the application and/or game. For example, some PCs have more or less RAM, or slow or fast CPUs, and slow or fast GPUs (if the PCs have a full GPU). Some games or applications take advantage of the full computing power of a given PC or console, and many games or applications do not take advantage of the full computing power of a given PC or console. If a user's choice of game or application does not reach the peak performance capability of the local PC or console, the user may be wasting money on the unused features of the PC or console. As far as the console is concerned, the console manufacturer may pay more than the money necessary to subsidize the console fee.

Another problem with the sale and enjoyment of video games involves allowing a user to watch other people play the game before proposing to purchase the game. There are several prior art methods for recording video games for replay at a later time. For example, U.S. Patent No. 5,558,339 teaches that game state information (including game controller actions) during "playing games" is recorded in a video game client computer (owned by the same or different users). This status information can be used to replay some or all of the game actions on a video game client computer (eg, a PC or console) at a later time. A significant disadvantage of this method is that for a user who views the recorded game, the user must have a video game client computer capable of playing the game and must be able to execute the video game application on the computer. The gameplay is exactly the same when replaying the recorded game state. In addition to this, the video game application must be written in such a way that there is no possible difference in execution between the recorded game and the played game.

For example, game graphics are typically calculated in a frame-by-frame manner. For many games, the game logic can sometimes calculate the graphics displayed in the next frame for a shorter or longer time than a frame time, depending on whether the scene is particularly complex, or whether there is a slowdown in execution. Other delays (for example, on a PC, another process that may be executing a CPU loop that occupies the game application). In this game, a "premise" frame can be calculated in a time that is slightly less than one frame time (ie, several CPU clock cycles or less). When the same scene is again calculated using the same game state information, it can easily take several CPU clock cycles that are more than one frame time (eg, if an internal CPU bus is slightly different from an external DRAM bus) In addition, it introduces a delay of several CPU cycle times, even if there is no large delay of another process of CPU time of a few milliseconds of self-occupying game processing). Therefore, when playing a game, the frame is calculated in two frame times instead of a single frame time. Some behaviors are based on the frequency at which the game calculates a new frame (eg, when the game samples input from the game controller). When playing a game, this deviation of different behaviors on the time reference does not affect the gameplay, but the bias can cause the game to play to produce a different result. For example, if the ballistics of the basketball is calculated at a steady rate of 60 fps, but the game controller input is sampled based on the calculated frame rate, the calculated frame rate can be 53 fps when recording the game. And 52 fps when replaying the game, this situation can cause a difference in the basketball to enter the basket, resulting in a different result. Therefore, using game state recording video games requires very careful game software design to ensure that replays using the same game state information produce exactly the same results.

Another prior art method for recording video games is to record only the video output of a PC or video game system (e.g., to a VCR, DVD recorder, or to a video capture card on a PC). The video can then be reversed and replayed, or the recorded video can be uploaded to the Internet, typically after compression. One disadvantage of this method is that when playing a 3D game sequence, the user is limited to viewing the sequence only from the point of view where the sequence was recorded. In other words, the user cannot change the viewing point of the scene.

In addition, when the compressed video of the recorded game sequence played on a home PC or game console can be obtained by other users via the Internet, even if the video is compressed in an instant manner, the video may not be immediately available. The compressed video is uploaded to the Internet. The reason is that many homes connected to the Internet in the world have very asymmetric broadband connections (for example, DSL and cable modems typically have much higher transmission bandwidths than the upload bandwidth). Compressed high resolution video sequences often have a higher bandwidth than the network's upload bandwidth capacity, making it impossible to upload the video sequences on the fly. Therefore, after playing the game sequence, there is a significant delay (perhaps a few minutes or even hours) before another user on the Internet can view the game. While this delay can be tolerated in certain situations (eg, watching a game played by a game player that occurred at a previous time), the delay eliminates the ability to watch a live game (eg, a basketball game by a champion player) or Has the ability to "immediately replay" when playing a live game.

Another prior art method allows a viewer with a television to watch a live video game and is only controlled by the television producer. Some television channels in the United States and other countries provide video game viewing channels in which television viewers can view a particular video game user on a video game channel (eg, a top player participating in the competition). This viewing is achieved by feeding the video output of the video game system (PC and/or console) to the video distribution and processing device of the television channel. This situation is the same as the television channel broadcasts a live basketball game in which several cameras provide live feeds at different angles around the basketball court. The television channel can then utilize its video/audio processing and effects devices to manipulate the output from various video game systems. For example, a television channel can overlay text indicating the status of different players on the video from a video game (like a TV channel can overlay text during a live basketball game), and the TV channel can be used from one. An audio dubbing of a commentator who can discuss the actions that take place during the game. In addition, the video game output can be combined with the camera recording video of the actual player of the game (eg, showing the player's emotional response to the game).

One problem with this approach is that the live video feed must be available in real time by the video distribution and processing device of the television channel to have the stimuli of live broadcast. However, as previously discussed, this is often not possible when the video game system is being executed in the home, especially if the portion of the broadcast includes live video from a camera that captures the real world video of the game player. In addition, in the case of a competition, the concern is that the home game player may modify the game and cheat as previously described. For these reasons, such video game broadcasts on television channels often arrange for players and video game systems to be grouped together in a common location (eg, gathered in a television studio or a competition venue) in which the television production equipment is available. Accept video feeds from multiple video game systems and possibly live cameras.

While such prior art video game television channels can provide a very exciting performance to a television viewer, the performance is an experience similar to a live sports game (eg, in terms of its actions in the video game world and in its In terms of action in the real world, video game players behave as "athletes", but such video game systems are often limited to situations where players are physically close to each other. Moreover, since the television channels are broadcast, each broadcast channel can only display one video stream selected by the producer of the television channel. Due to these limitations and the high cost of broadcast time, production equipment and production staff, these TV channels usually only show top players who participate in top competitions.

In addition, a given television channel that broadcasts one full video image of a video game to all television viewers only displays one video game at a time. This strictly limits the choice of TV viewers. For example, a television viewer may not be interested in a game displayed at a given time. Another viewer may only be interested in watching the television channel not giving a particular player a particular gameplay game at a given time. In other cases, the viewer may only be interested in watching how a seasoned player handles a particular level in a game. In addition, other viewers may wish to control the viewing point of the video game (the viewing point is different from the viewing point selected by the production team) and the like. In short, even though a number of different television channels are available, television viewers may have numerous preferences for viewing video games, which may not be accommodated by a particular broadcast of the television network. For all of the foregoing reasons, prior art video game television channels have significant limitations in presenting video games to television viewers.

Another disadvantage of prior art video game systems and application software systems is that they are complex and often subject to errors, corruption and/or unwanted and inappropriate behavior (collectively referred to as "program errors"). While games and applications typically go through a debugging and tuning process (often referred to as "software quality assurance" or SQA) before release, it is almost unchanged that once a game or application is released to a wide audience in the field, A program error will suddenly appear. Unfortunately, it is difficult for software developers to identify and track many of these program errors after the release. Software developers may have difficulty noticeing program errors. Even if the developers know of a program error, there may only be a limited amount of information available to identify the cause of the program error. For example, the user can call a game developer's customer service line and leave a message stating that when the game is played, the screen begins to flash, then becomes solid blue and the PC freezes. This message provides the SQA team with very little information useful in tracking a bug. Some games or applications that are connected online can sometimes provide more information in specific situations. For example, a "watchdog" handler can sometimes be used to monitor the "damage" of a game or application. The watchdog handler collects statistics about the state of the game or application handler at the time of corruption (eg, the state of the stack, the state of the memory usage, how much the game or application has been performed, etc.) and then via the Internet The road uploads the information to the SQA team. However, in a complex game or application, decrypting this information to accurately determine what the user is doing while it is damaging may take a long time. Even so, it may not be possible to determine which sequence of events caused the damage.

Yet another problem associated with PCs and game consoles is that they suffer from service problems that are very inconvenient for consumers. Service issues also affect the manufacturer of the PC or game console because it typically requires a special box to be sent to safely transport the damaged PC or console, and if the PC or console has a service guarantee, repair costs are incurred. Game or application software publishers may also be affected by sales losses (or online service usage) due to the PC and/or console being repaired.

Figure 1 illustrates a prior art video game system, such as a Sony Playstation 3, Microsoft Xbox 360 , Nintendo Wii ^TM , Windows-based PC or Apple Macintosh. Each of these systems includes a central processing unit (CPU) for executing code (usually a graphics processing unit (GPU) for performing advanced graphics operations), and for use with external devices and users. Various forms of input/output (I/O) for communication. For the sake of simplicity, these components are grouped together to be presented as a single unit 100. Also shown is a prior art video game system of FIG. 1 including: an optical media drive 104 (eg, a DVD-ROM drive); a hard drive 103 for storing video game code and data; Playing a multiplayer game network connection 105 for downloading games, patches, demonstrations or other media; a random access memory (RAM) 101 for storing code currently executed by the CPU/GPU 100; A game controller 106 for receiving an input command from a user during game play; and a display device 102 (eg, a standard definition television (SDTV)/high definition television (HDTV) or a computer monitor).

The prior art system shown in Figure 1 suffers from several limitations. First, the optical disk drive 104 and the hard disk drive 103 tend to have a much slower access speed than the access speed of the RAM 101. In practice, when working directly via the RAM 101, the CPU/GPU 100 can process polygons per second much more than what can be processed when reading code and data directly from the hard disk drive 103 or the optical disk drive 104. Because RAM 101 typically has a higher frequency bandwidth and does not suffer from the relatively long search delay of the disc mechanism. However, only a limited amount of RAM (e.g., 256-512 million bytes) is provided in such prior art systems. Therefore, a "loading..." sequence is often required, in which the RAM 101 periodically fills in the data of a scene below the video game.

Some systems attempt to overlap the loading of code while playing the game, but this overlap can only be done when there is a known sequence of events (for example, if a car is traveling along a road, the asymptotic buildings beside the road) The geometry can be loaded while the car is driving). For complex and/or fast scene changes, this type of overlap usually does not work. For example, if the user is in combat and the RAM 101 is completely filled with information indicating objects that are visible at the time, if the user quickly moves the field of view to the left to view that it is not currently loaded in the RAM 101. The object will create a discontinuity in motion because there is not enough time to load the new object from the hard disk drive 103 or optical media 104 into the RAM 101.

Another problem with the system of Figure 1 arises due to the limitations of the storage capacity of the hard disk drive 103 and optical media 104. Although disk storage devices having a relatively large storage capacity (eg, 50 billion bytes or more) can be fabricated, the disk storage devices are still not specific to those encountered in current video games. The situation provides sufficient storage capacity. For example, as previously mentioned, a soccer video game may allow a user to select among many teams, players, and sports fields worldwide. For each team, every player and every stadium, many texture maps and environment maps are needed to characterize the 3D surface in the game world (for example, each team has a unique jersey and each jersey needs a unique texture) Figure).

One technique for solving this latter problem is that once the user selects a texture map and an environment map for the game, the texture map and the environment map are pre-calculated. This technique may involve many computationally intensive processing procedures, including decompressing images, 3D graphics, shading, organizing data structures, and the like. As a result, the user may feel the delay when the video game performs such calculations. One way to reduce this delay is in principle to perform all such calculations at the time of the initial development of the game, including each arrangement of the team, the roster of players and the playing field. The released version of the game may thus include all of this pre-processed material stored on one or more servers on the optical media 104 or the Internet, and after the user makes a selection, a given via the Internet The selected pre-processing data for the team, player roster, and sports field selection is downloaded to the hard disk drive 103. However, as a practical matter, this preloaded data for each permutation possible in a game play can easily be a piece of data for a number of terabytes, which far exceeds the capacity of today's optical media devices. In addition, a given team, player roster, and sports field selection information can easily be hundreds of millions of bytes of information or more. Under a home network connection of 10 Mbps, for example, the time taken to download the data via the network connection 105 will be longer than the time taken by the local end to calculate the data.

Thus, the prior art gaming architecture shown in Figure 1 exposes the user to significant delays between major scene transitions of complex games.

Another problem with prior art methods such as the method shown in Figure 1 is that video games tend to become more advanced and require higher CPU/GPU processing power over the years. Therefore, even if an unlimited amount of RAM is assumed, the video game hardware requirements will exceed the peak level of processing power available in such systems. As a result, users are required to upgrade their game hardware every few years to keep up with current technology (or play newer games at lower quality levels). One of the trends in video games that are constantly changing is that video game players for home use are generally economically inefficient because their fees are usually determined by the demand for the most powerful games they can support. For example, the XBox 360 can be used to play a game like "Gears of War", which requires a high-performance CPU, GPU and hundreds of millions of bytes of RAM, or XBox 360 can be used to play Pac Man, a 20 The game of the 1970s required only a few thousand bytes of RAM and a very low-performance CPU. In fact, the XBox 360 has the computing power to simultaneously host many simultaneous Pac Man games.

Video game consoles are usually turned off most of the week. According to Nielsen Entertainment Research, a player in the role of 13 years old and older in July 2006, the average player spends 14 hours per week or 12% of the total time of the week on the video game console. This means that the average video game console is idle 88% of the time, which is an inefficient use of expensive resources. This is especially true if the video game console is often subsidized by the manufacturer to lower the purchase price (expected to earn back subsidies through royalties purchased by future video game software).

The video game console also generates fees associated with almost any consumer electronic device. For example, the electronics and mechanisms of such systems must be housed in a housing. The manufacturer must provide a service guarantee. Retailers selling the system must make a profit from the sale of the system and/or from the sale of video game software. All of these factors increase the cost of the video game console, which must be subsidized by the manufacturer, passed on to the consumer, or both.

In addition, piracy is a major problem in the video game industry. In essence, the security mechanisms used on each of the major video game systems have "collapsed" over the years, resulting in unauthorized copying of video games. For example, the Xbox 360 security system crashed in July 2006, and users can now download illegal copies online. Downloadable games (for example, games on PC or Mac) are particularly vulnerable to piracy. In some areas of the world where piracy is not well regulated, there is virtually no viable market for independent video game software, because users can buy pirated copies at a fraction of the cost, just as easily as purchasing a legal copy. In addition, in many parts of the world, the cost of the game console accounts for a high percentage of revenue, so that even if piracy is controlled, few people can afford the current state of the art gaming system.

In addition, the old game market has reduced the revenue of the video game industry. When a user is tired of a game, he can sell the game to a store that resells the game to other users. This unauthorized but common practice has resulted in a significant reduction in the revenue of game publishers. Similarly, when a platform transition occurs every few years, there is typically a 50% reduction in sales. This is because the user stops buying games for older platforms when they know that a newer version of the platform is about to be released (for example, when the Playstation 3 is about to be released, the user stops purchasing the Playstation 2 game). Taken together, lost sales and increased development costs associated with the new platform can have a particularly significant negative impact on the game developer's revenue.

The new game console is also very expensive. The retail prices for the Xbox 360, Nintendo Wii and Sony Playstation 3 are all in the hundreds of dollars. High-performance PC gaming systems can cost as much as $8,000. This is a significant investment by one of the users, especially considering the fact that the hardware becomes useless after a few years and many systems are purchased for children.

One solution to the foregoing problem is an online game in which game code and data are hosted on a server and transmitted as a compressed video and audio streamed over a digital broadband network to a client machine as needed. Some companies, such as Finnish G-Cluster (now a subsidiary of SOFTBANK Broadmedia in Japan), currently offer these services online. Similar gaming services have become available in regional networks, such as regional networks within hotels and provided by DSL and cable providers. One of the major drawbacks of such systems is the problem of latency (i.e., the time it takes for the signal to travel to the game server and travel from the game server), which is typically located at the "head-end" of the industry. A fast motion video game (also known as a "twitch" video game) requires very low latency between when the user performs an action with the game controller and when the display screen is updated to show the result of the user action. Low latency is required to give the user the feeling that the game is "immediately" responsive. Depending on the type of game and the skill level of the user, the user can be satisfied with different delay intervals. For example, a 100 millisecond delay is a slow casual game (such as backgammon) or a slow motion role-playing game that can be tolerated, but in a fast action game, a delay of more than 70 milliseconds or 80 milliseconds can cause the user to play. The performance is poor and therefore unacceptable. For example, in a game that requires fast response time, the accuracy drops dramatically as the delay increases from 50 milliseconds to 100 milliseconds.

When installing a game or application server in a nearby controlled network environment or into a network environment where the user's network path is predictable and/or tolerant of peak bandwidth, the maximum latency is In terms of the consistency of the delay, it is easy to control the delay (for example, the user observes the steady motion of the digital video from the Internet stream). This control can be achieved between the cable TV network headend and the cable subscriber home, or from the DSL Directorate to the DSL subscriber home, or in a server or a user's business office area network (LAN) environment. grade. In addition, it is possible to obtain a specially graded point-to-point private connection between enterprises with guaranteed bandwidth and latency. However, in a game center connected to the server network of the general Internet and then the compressed video stream is streamed to the user's game or application system via a broadband connection, the delay is caused by many factors, resulting in Strict restrictions on the deployment of prior art systems.

In a typical broadband connected home, the user may have a DSL or cable modem for broadband service. These broadband services typically cause up to 25 milliseconds of round-trip delay (and sometimes higher) between the user's home and the general Internet. In addition, there is a round trip delay caused by the delivery of data via the Internet to a server center. The delay through the Internet changes based on the path of the given data and the delay caused by the delivery of the data. In addition to routing delays, round-trip delays are also caused by the speed of light passing through the fibers that interconnect most of the Internet. For example, due to the speed of light passing through the fiber and other additional burdens, a round trip delay of approximately 22 milliseconds per 1000 turns is caused.

Additional delays can occur due to the data rate of the data streamed through the Internet. For example, if a user has a DSL service that is sold as a "6 Mbps DSL service," in practice, the user may get a transmission volume of less than 5 Mbps, and is likely to see: connection attribution. Periodically degraded by various factors, such as the congestion of a Digital Subscriber Line Access Multiplexer (DSLAM) during peak or ingress time. A similar problem may arise in the case where the loop is concomitant in a shared coaxial cable through a neighborhood or in a place other than in the network of the cable modem system, and is sold as a "6 Mbps cable modem service". The data rate of one of the cable modems is reduced to much less than the data rate. If the data packet of the stable rate of 4 Mbps is unidirectionally streamed from a server center via the connection in the User Datagram Protocol (UDP) format, if all goes well, the data packets will pass without causing Additional delay, but if there is congestion (or other obstacles) and only 3.5 Mbps can be used to stream data to the user, then in the typical case, the packet will be discarded, resulting in data loss, or the packet will be at the congestion point The queue is queued until the packets can be sent, thereby introducing additional delays. Different congestion points have different queue capacities for packets that are delayed, so in some cases, packets that cannot pass through the congestion are immediately discarded. In other cases, several million bits of data are queued and eventually sent. However, in almost all cases, the queue at the congestion point has a capacity limit, and once it exceeds these limits, the queue will overflow and the packet will be discarded. Therefore, in order to avoid causing additional delays (or worse, packet loss), data rate capacity beyond the game or application server to the user must be avoided.

The time required for the server to compress the video and the client device to decompress the video also causes a delay. When a video game executed on a server is calculating the next frame to be displayed, a delay is further caused. Currently available video compression algorithms suffer from high data rates or high latency. For example, Motion JPEG is a frame-only lossy compression algorithm characterized by low latency. Each frame of the video is compressed independently of each other frame of the video. When a user device receives a frame of compressed motion JPEG video, the client device can immediately decompress the frame and display the frame, resulting in very low latency. However, because each frame is compressed separately, the algorithm cannot take advantage of the similarity between successive frames, and as a result, only the in-frame video compression algorithm suffers from extremely high data rates. For example, a 60 fps (frames per second) 640x480 motion JPEG video may require 40 Mbps (megabits per second) or higher. In many broadband applications (and undoubtedly for most consumers, Internet-based applications), such high-resolution data rates are prohibitively high. In addition, because each frame is compressed independently, artifacts that may be caused by lossy compression in the frame may appear at different locations in the continuous frame. This can result in a moving visual artifact being presented to the viewer after the video is decompressed.

Other compression algorithms such as MPEG2, H.264, or VC9 from Microsoft Corporation can achieve high compression ratios when used in prior art configurations, but high compression ratios are at the expense of high latency. These algorithms utilize compression between frames and within frames. These algorithms periodically perform an intra-frame only compression on a frame. This frame is called a key frame (often referred to as an "I" frame). These algorithms then typically compare the I frame to the previous and subsequent frames. The previous frame and subsequent frames are not compressed independently. The algorithm determines the changes that have occurred between the I frame and the previous frame and subsequent frames in the image, and then stores the changes as "B". Frame (change before I frame) and "P" frame (change after I frame). This results in a much lower data rate than compression within the frame only. However, this algorithm is usually implemented at the expense of higher latency. The I frame is much larger (often 10 times) than the B or P frame, and as a result, the time taken to transmit the I frame at a given data rate increases proportionally.

For example, consider the following situation: the size of the I frame is 10 times that of the B frame and the P frame, and there are 29 B frames + 30 P frames = 59 intermediate images in each single I inner frame. The box, or each "Group of Frames" (GOP), has a total of 60 frames. Therefore, at 60 fps, there is one 60-frame GOP per second. Assume that the transmission channel has a maximum data rate of 2 Mbps. In order to achieve the highest quality video in the channel, the compression algorithm can generate a 2 Mbps data stream, and given the above ratio, this can result in 2 million bits (Mb)/(59+10) per inner frame. ) = 30,394 bits and 303, 935 bits per I frame. When a compressed video stream is received by a decompression algorithm, in order to play the video stably, each frame must be decompressed and displayed at regular intervals (eg, 60 fps). To achieve this result, if any frame is subjected to a transmission delay, all frames must be delayed by at least that delay, so the worst case frame delay will define the delay for each video frame. The I frames introduce the longest transmission delay due to their maximum, and the entire I frame will have to be received before decompressing and displaying the I frame (or any intermediate frame that depends on the I frame). If the channel data rate is 2 Mbps, transmitting an I frame will take 303,935/2Mb=145 milliseconds.

Due to the larger size of the I frame relative to the average size of a frame, an inter-frame video compression system that uses a large percentage of the bandwidth of the transmission channel as described above will experience long delays. Or, in other words, although the prior art inter-frame compression algorithm achieves an average per-frame data rate that is lower than the in-frame compression algorithm (eg, 2 Mbps versus 40 Mbps), the algorithms are due to large I The frame still suffers from a high peak per frame data rate (eg, 303, 935 * 60 = 18.2 Mbps). It should be remembered that, despite this, the above analysis assumes that the P and B frames are much smaller than the I frame. While this is often true, it is not true for frames that have high image complexity that is unrelated to previous frames, high motion, or scene changes. In these cases, the P or B frame can become as large as the I frame (if a P or B frame becomes larger than an I frame, a perfect compression algorithm will usually be "imposed". An I frame and replace the P or B frame with an I frame). Therefore, the data rate peak of the I frame size can appear in a digital video stream at any time. Therefore, for compressed video, when the average video data rate is close to the data rate capacity of the transmission channel (this is a common situation, given the high data rate requirement for video), I frame or large P or B frame A high peak data rate results in a high frame delay.

Of course, the above discussion only characterizes the compression algorithm delay caused by the large B, P or I frame in the GOP. If you use the B frame, the delay will be higher. The reason is that all B frames and the I frames after the B frame must be received before a B frame can be displayed. Therefore, in a group of image groups (GOP) such as BBBBBIPPPPPBBBBBIPPPPP (with 5 B frames before each I frame), before receiving the subsequent B frame and I frame, the first B picture The box cannot be displayed by the video decompressor. Therefore, if the video is being streamed at 60 fps (ie, 16.67 ms/frame), it will take 16.67*6=100 milliseconds before the first B frame can be decompressed, regardless of the channel bandwidth. To receive five B frames and the I frame, and this has only 5 B frames. A compressed video sequence with 30 B-frames is quite common. Moreover, in the low channel bandwidth (such as 2 Mbps), the delay effect caused by the size of the I frame is mainly added to the delay caused by the arrival of the waiting B frame. Therefore, on a 2 Mbps channel, for many B-frames, it is easy to use a prior art video compression technique that exceeds a delay of 500 milliseconds or a longer delay. If the B-frame is not used (at the expense of a lower compression ratio for a given quality level), the B-frame delay is not caused, but still causes the delay caused by the above-mentioned peak frame size.

The nature of many video games has exacerbated the problem. The video compression algorithm utilizing the GOP structure described above has been largely optimized for use with live video or animated materials intended for passive viewing. Usually, the camera (whether it is a real camera or a virtual camera in the case of a computer-generated animation) and the scene are relatively stable, simply because if the camera or scene moves around too trembling, watch the video or movie material (a) Often unpleasant, and (b) if you are watching, the viewer usually does not closely watch the action when the camera suddenly trembles (for example, if the camera is shocked when shooting a child to blow out the candle on the birthday cake and suddenly trembles away from the cake And to restore, the audience is usually concentrated on the children and the cake, and ignores the brief interruption when the camera suddenly moves). In the case of a video interview or a video conference call, the camera can be held in a fixed position and does not move at all, resulting in minimal data spikes at all. However, 3D high-performance video games are characterized by constant motion (for example, considering 3D racing, where the entire frame moves fast during the duration of the game, or considers first-person shooters, where the virtual camera always moves around with shaking) . Such video games can produce a sequence of frames with large and frequent peaks, where the user may have to clearly see what happens during their sudden movements. Therefore, compression artifacts are less tolerable in 3D high-performance video games. Therefore, the video output of many video games produces a compressed video stream with extremely high and frequent peaks due to its nature.

If the user of the fast motion video game can hardly tolerate high latency and gives the full delay or more, there have been limitations on the video game hosted by the server for streaming video over the Internet. In addition, if an application is hosted on a general Internet and the application is streaming video, users of a highly interactive application will be subject to similar restrictions. These services require a network configuration in which the escrow server is set directly in a commercial environment (in the case of cable broadband) or in the case of a central office (digital subscriber line (DSL)) or in a LAN ( Or on a special graded private connection), the route and distance from the client device to the server are controlled to minimize delay and can accommodate peaks without causing delays. LANs with sufficient bandwidth (usually with a rate of 100 Mbps-1 Gbps) and leased lines typically support peak bandwidth requirements (for example, a peak bandwidth of 18 Mbps is a fraction of a 100 Mbps LAN capacity).

If a special facility is formed, the residential broadband infrastructure can also accommodate peak bandwidth requirements. For example, on a cable television system, a digital video-specific bandwidth can be given that can handle peaks such as large I-frames. Moreover, on a DSL system, a higher speed DSL modem can be provided to allow for high peaks or to provide a special hierarchical connection that can handle a higher data rate. However, conventional cable modems and DSL infrastructure attached to the general Internet are far from meeting the peak bandwidth requirements of compressed video. Therefore, online services (which host video games or applications in a server center that is far from the client device, and then stream compressed video output over the Internet via a conventional residential broadband connection) suffer significant delays and Peak bandwidth limits, especially for games and applications that require very low latency (for example, interactive action games for first-person shooters and other multi-users, or applications that require a quick response time).

The present invention will be more fully understood from the following description of the accompanying drawings. Used for explanation and understanding.

In the following description, specific details such as device types, system configurations, communication methods, and the like are set forth to provide a thorough understanding of the invention. However, it will be appreciated by those skilled in the art that the specific embodiments are not necessarily required to practice the described embodiments.

2a-2b provide a high-level architecture of two embodiments in which video games and software applications are hosted by the colocation service 210 and hosted by the user device 205 at the user premises 211 via the Internet 206 (or other public) Or private network) access by subscription service (please note that "user location" means the user's location, including outdoor use when using mobile devices). The client device 205 can be a general purpose computer such as a Microsoft Windows based or Linux based PC or an Apple, Inc. Macintosh computer (with one of the wired or wireless connections to the Internet) having an internal or external display device 222, or The client device can be a dedicated client device, such as a video converter that outputs video and audio to a monitor or television 222 (with wired or wireless connection to the Internet), or the user device can act The device, which may have a wireless connection to the Internet.

Any of these devices may have its own user input device (eg, keyboard, button, touch screen, trackpad or inertial sensor, video capture camera, and/or motion tracking camera, etc.), or These devices may use an external input device 221 that is wired or wirelessly connected (eg, a keyboard, mouse, game controller, inertial sensing stick, video capture camera, and/or motion tracking camera, etc.). As described in more detail below, the colocation service 210 includes servers of various performance levels, including servers with high performance CPU/GPU processing capabilities. During play of the game on the colocation service 210 or the application on the colocation service 210, the home or office client device 205 receives keyboard and/or controller input from the user, and then, the user device via the user device The Internet 206 transmits the controller input to the colocation service 210, which acts as a continuation frame (a sequence of video images) in response to executing the game code and generating a video output of the game or application software ( For example, if the user presses a button that moves to the right of one of the characters on the screen, the game program will then form a sequence of video images showing that the character is moving rightward. The sequence of video images is then compressed using a low latency video compressor, and the colocation service 210 then transmits the low latency video stream via the internet 206. The home or office client device then decodes the compressed video stream and reproduces the decompressed video image on a monitor or TV. Therefore, the computational and graphical hardware requirements of the client device 205 are significantly reduced. The client 205 only has to have the ability to forward the keyboard/controller input to the Internet 206 and decode and decompress a compressed video stream received from the Internet 206. Essentially, any personal computer can now This processing is performed in software on the CPU (for example, an Intel Corporation Core Duo CPU executing at approximately 2 GHz can decompress 720p HDTV encoded using a compressor such as H.264 and Windows Media VC9). Moreover, in the case of any client device, the dedicated chip can also perform these standard video decompressions at a much lower cost and much less power consumption than a general purpose CPU (such as a CPU that may be required by a modern PC). . It is worth noting that in order to perform the functions of forwarding controller input and decompressing video, the home client device 205 does not require any special graphics processing unit (GPU), optical disk drive or hard disk drive, such as the prior art video game shown in FIG. system.

As games and application software become more complex and more realistic, games and applications may require higher performance CPUs, GPUs, more RAM, and larger and faster drives, and may continue to upgrade. The host hosts the computing power of the service 210, but the end user does not need to update the home or office client platform 205 because the platform handles the display resolution and frame rate with a given video decompression algorithm. Demand will remain constant. Therefore, the hardware limitations and compatibility issues seen today do not exist in the systems illustrated in Figures 2a-2b.

In addition, because the game and application software is only executed in the server in the colocation service 210, there will never be a copy of the game or application software in the user's home or office (in the form of optical media, or as Downloaded Software) (Unless otherwise limited, the "office" used herein shall include any non-residential environment, including, for example, classrooms). This significantly reduces the likelihood that a game or application software will be illegally copied (pirated) and may also reduce the likelihood that a valuable database that can be used by a game or application will be pirated. In fact, if you need a dedicated server that is not suitable for home or office use (for example, you need a very expensive, large or noisy device) to play games or application software, even if you get a pirated copy of the game or application software, You cannot operate it at home or in the office.

In one embodiment, the colocation service 210 provides a software development tool to a game or application software developer who designs a video game (which is typically referred to as a software development company, a game or movie studio, or a game or application software publisher). 220, enabling a game or application software developer to design a game that can be executed on the colocation service 210. These tools allow developers to take advantage of the features of colocation services that are typically not available in a standalone PC or game console (for example, quick access to very large databases of complex geometries (unless otherwise limited, otherwise geometry) Used herein to refer to polygons, textures, rigging, lighting, behavior, and other components and parameters that define 3D datasets)).

There may be different enterprise models under this architecture. Under a model, the colocation service 210 charges a subscription fee from the end user and pays royalties to the developer 220, as shown in Figure 2a. In an alternate implementation, as shown in FIG. 2b, the developer 220 charges the subscription fee directly from the user and pays for the colocation service 210 for the hosted game or application content. These basic principles are not limited to any particular enterprise model used to provide online game or application hosting.

Compressed video features

As discussed previously, one of the significant issues of providing video game services or application software services in an online manner is the latency issue. A delay of 70 milliseconds to 80 milliseconds (from the time the user activates an input device to the time the response is displayed on the display device) is the upper limit of the game and application that requires a quick response time. However, due to many practical and physical constraints, it is extremely difficult to achieve this delay in the case of the architecture shown in Figures 2a and 2b.

As indicated in Figure 3, when a user subscribes to an internet service, the connection is typically priced at a nominal maximum data rate 301 to the user's home or office. Depending on the provider's strategy and routing device capabilities, this maximum data rate may be more or less strictly enforced, but for one of many different reasons, the actual available data rate is typically lower. For example, there may be too many network traffic on the DSL headquarters or on the local cable modem loop, or there may be noise on the cable that causes the packet to be dropped, or the provider can determine the maximum monthly size of each user. Yuan. Currently, the maximum downlink data rate for cable and DSL services is typically between a few hundred kilobits per second (Kbps) and 30 Mbps. Honeycomb services are usually limited to hundreds of Kbps. However, the speed of broadband services and the number of users subscribed to broadband services will increase dramatically over time. Currently, some analysts estimate that 33% of US broadband users have a downlink data rate of 2 Mbps or higher. For example, some analysts predict that by 2010, more than 85% of US broadband users will have a data rate of 2 Mbps or higher.

As indicated in Figure 3, the actual available maximum data rate 302 may fluctuate over time. Therefore, in the case of low latency online games or application software, it is sometimes difficult to predict the actual available data rate for a particular video stream. The data rate required to maintain a given quality level for a given amount of scene complexity and motion at a given resolution and a given number of frames per second (fps) (eg, 640 x 480 at 60 fps) 303 exceeds the actual available maximum data rate 302 (as indicated by the peak in Figure 3), and several problems may arise. For example, some Internet services will only discard packets, resulting in loss of data and distortion/loss of images on the user's video screen. Other services will temporarily buffer additional packets (ie, queued) and provide the packets to the client at the available data rate, resulting in increased latency, which is an unacceptable result for many video games and applications. Finally, some Internet service providers see this increase in data rate as a malicious attack, such as blocking service attacks (a well-known technology used by hackers to disable network connections) and disrupting the user's Internet connection. A specified time period. Thus, the embodiments described herein seek to ensure that the required data rate for a video game does not exceed the maximum available data rate.

Hosted service architecture

Figure 4a illustrates the architecture of a colocation service 210 in accordance with an embodiment. The colocation service 210 can be located in a single server center or can be distributed across a plurality of server centers (to provide users with lower latency connections (these users to a particular server center have more than other server centers) Lower latency path) provides load balancing between users and provides redundancy in the event of one or more server center failures). The colocation service 210 may ultimately include hundreds of thousands or even millions of servers 402 that serve an enormous pool of users. The colocation service control system 401 provides overall control over the colocation service 210 and directs routers, servers, video compression systems, billing and account processing systems, and the like. In one embodiment, the colocation service control system 401 is implemented on a decentralized Linux-based processing system that is coupled to an independent magnetic library for storing user information, server information, and system statistics. Redundant Array of Disks (RAID). In the foregoing description, various actions performed by the colocation service 210 have been initiated and controlled by the colocation service control system 401 (unless such actions are attributed to other specific systems).

The colocation service 210 includes a number of servers 402, such as those currently available from Intel, IBM, and Hewlett Packard, and others. Alternatively, the servers 402 can be assembled in one of the custom configuration configurations, or can ultimately be integrated such that the entire server is implemented as a single wafer. Although this figure shows a small number of servers 402 for illustrative reasons, in actual deployments there may be as few as one server 402, or as many as millions of servers 402, or more servers. Server 402 can all be configured in the same way (as one of the configuration parameters, with the same CPU type and performance; with or without GPU, and with GPU, with the same GPU type and performance; The same number of CPUs and GPUs; with the same amount of RAM and the same RAM type/speed; and with the same RAM configuration), or various subsets of servers 402 can have the same configuration (eg, 25% of the servers can In a specific way, 50% of the servers are configured in a different way, and 25% of the servers are configured in another way, or each server 402 can be different.

In one embodiment, the server 402 is free of magnetic disks, that is, does not have its own native mass storage device (which is an optical or magnetic storage device, or a semiconductor-based storage device such as a flash memory, or Other large-capacity storage components of the servo-like function, each server accesses the shared mass storage via a fast backplane or internet connection. In one embodiment, the quick connect is a storage area network (SAN) 403 connected to a series of independent redundant arrays of RAID (RAID) 405, and the connections between the devices use a billion bit Ethernet. To implement. As is known to those skilled in the art, the SAN 403 can be used to combine many RAID arrays 405, resulting in extremely high frequency bandwidths that are close to or possibly exceed the bandwidth available from the current gaming console and the RAM used in the PC. Thus, while RAID arrays based on rotating media (such as magnetic media) often have significant seek-time access latency, semiconductor memory-based RAID arrays can be implemented with much lower access latency. In another configuration, some or all of the servers 402 provide some or all of their own mass storage at the local end. For example, the server 402 can store frequently accessed information (such as a copy of its operating system and a video game or application) on a low-latency local flash memory-based storage, but The server can use a SAN to access a rotating media-based RAID array 405 with a higher seek latency at a lower frequency to access a large database of geometry or game state information.

Additionally, in an embodiment, the colocation service 210 uses low latency video compression logic 404, which will be described in detail below. Video compression logic 404 can be implemented in software, hardware, or any combination thereof (a specific embodiment of the logic will be described below). Video compression logic 404 includes logic for compressing audio as well as visual material.

In operation, when a video game or application is played at the user location 211 via a keyboard, mouse, game controller or other input device 421, the control signal logic 413 on the client 415 will indicate that the user actuated Control signals 406a-b (typically in the form of UDP packets) of button presses (and other types of user inputs) are transmitted to the colocation service 210. Control signals from a given user are routed to the appropriate server (or servers if multiple servers are responsive to the user's input device) 402. Control signal 406a may be routed to server 402 via the SAN as illustrated in Figure 4a. Additionally or alternatively, control signal 406b may be routed directly to server 402 via a colocation service network (e.g., an Ethernet based local area network). Regardless of the manner in which the control signals are transmitted, the or the servers execute the game or application software in response to the control signals 406a-b. Although not illustrated in FIG. 4a, various network connectivity components, such as firewalls and/or gateways, may be at the edge of the colocation service 210 (eg, between the colocation service 210 and the Internet 410) and/or The incoming and outgoing traffic is processed at the edge of the user premises 211 between the Internet 410 and the home or office client 415. The graphics and audio output (i.e., the new sequence of video images) of the executed game or application software is provided to low latency video compression logic 404, which is based on low latency video compression techniques (such as video compression as described herein). Technology) compressing the sequences of video images and compressing the video stream via the Internet 410 (or as described below, via an optimized high speed network service bypassing the general Internet) (typically having The compressed or uncompressed audio is transmitted back to the client 415. The low latency video decompression logic 412 on the client 415 then decompresses the video stream and audio stream and reproduces the decompressed video stream on the display device 422 and typically plays the decompressed audio stream. Alternatively, the audio may be played by a speaker separate from display device 422 or not played at all. Note that although the input device 421 and the display device 422 are shown as separate devices in Figures 2a and 2b, the devices can be integrated into a client device such as a portable computer or mobile device.

The home or office client 415 (described previously as home or office client 205 in Figures 2a and 2b) can be a very inexpensive and low power device (with very limited computational or graphical performance) and is likely to be extremely limited. The local mass storage device does not have a local mass storage device. In contrast, each server 402 coupled to a SAN 403 and a plurality of RAIDs 405 can be an exceptionally high performance computing system and, in effect, can be used in a cooperative manner in a parallel processing configuration. With the server, there is almost no limit to the amount of computing and graphics processing power available. Therefore, due to the low latency video compression 404 and the low latency video decompression 412, the user can perceive that the computing power of the server 402 is being provided to the user. When the user presses a button on the input device 421, the image on the display 422 is updated in response to the button press, and there is no significant delay, as if the game or application software was executing at the local end. Thus, in the case of a home or office client 415 that is a very low performance computer or an inexpensive chip that implements low latency video decompression and control signal logic 413, the user effectively has arbitrary calculations from a remote location. Ability, which seems to be available at the local end. This gives users the ability to play the most advanced processor-intensive (usually new) video games and the most productive applications.

Figure 4c shows a very basic and inexpensive home or office client device 465. This device is an embodiment of the home or office client 415 from Figures 4a and 4b. The device is approximately 2 inches long. The device has an Ethernet jack 462 interfaced with an Ethernet cable (Ethernet over Ethernet) (PoE), which derives its power and connectivity to the Internet from the cable. . It is capable of performing NAT in a network that supports Network Address Translation (NAT). In an office environment, many new Ethernet switches have PoE and deliver PoE directly to the Ethernet jack in the office. In this case, only one Ethernet cable from the wall jack to the customer terminal 465 is required. If you can use an Ethernet connection without a power supply (for example, in a home with a DSL or cable modem without PoE), you can use inexpensive wall "bricks" (ie, power supplies) that accept one There is no power Ethernet cable and the Ethernet with PoE is output.

The client 465 includes control signal logic 413 (FIG. 4a) coupled to a Bluetooth wireless interface, the Bluetooth wireless interface and a Bluetooth input device 479 (such as a keyboard, mouse, game controller, and/or microphone). / or headset) interface. In addition, an embodiment of the client 465 can output video at 120 fps and is coupled to the display device 468 and a pair of shutter glasses 466, the display device capable of supporting 120 fps video and signals (usually via infrared), and the shutter The glasses alternately mask one eye with each successive frame and then obscure the other eye. The effect that the user perceives is a stereoscopic 3D image of the "bounce" display. One such display device 468 that supports this operation is the Samsung HL-T5076S. Since the video stream for each eye is separate, in one embodiment, the colocation service 210 compresses two independent video streams, the frames are interleaved in time, and the frames are used as users. Decompressed by two independent decompression handlers in terminal 465.

Client 465 also includes low latency video decompression logic 412 that decompresses incoming video and audio and outputs via High Definition Multimedia Interface (HDMI) connector 463, which is plugged into standard definition television (SDTV) or high. In the definition television (HDTV) 468, video and audio are provided for the TV, or inserted into the monitor 468 supporting HDMI. If the user's monitor 468 does not support HDMI, an HDMI-to-DVI (digital video interface) can be used, but the audio will be lost. According to the HDMI standard, display capabilities (eg, supported resolution, frame rate) 464 are communicated from display device 468, and then this information is passed back to the colocation service 210 via the internet connection 462, thus, the host generation The pipe service can compress the video in a format suitable for the display device.

Figure 4d shows a home or office client device 475 that is identical to the home or office client device 465 shown in Figure 4c, except that the client device has more external interfaces. In addition, the client 475 can be powered by PoE, or it can be used with an external power supply adapter (not shown) that is plugged into the wall. Using the client 475 USB input, the video camera 477 provides the compressed video to the client 475, which is uploaded by the client 475 to the colocation service 210 for use as described below. A low latency compressor utilizing the compression technique described below is built into the camera 477.

In addition to having an Ethernet connector for Internet connection, the client 475 also has an 802.11g wireless interface to one of the Internet. These interfaces are capable of using NAT in a network that supports NAT.

In addition, in addition to having an HDMI connector for outputting video and audio, the client 475 also has a dual-link DVI-I connector that includes an analog output (and a standard adapter will provide a VGA output) ). The client also has an analog output for composite video and S video.

For audio, the client 475 has a left/right analog stereo RCA jack, while for digital audio output, the client 475 has a TOSLINK output.

In addition to the Bluetooth wireless interface of one of the input devices 479, the client also has a USB jack for interfacing to the input device.

Figure 4e shows an embodiment of the internal architecture of the client 465. All or some of the devices shown in the figures can be implemented in a programmable logic array, implemented in a custom ASIC, or implemented in custom designs or off-the-shelf discrete devices.

Ethernet 497 with PoE is attached to the Ethernet interface 481. Power source 499 is derived from Ethernet 497 with PoE and is connected to other devices in client 465. Bus 480 is a common bus for communication between devices.

Execute a control CPU 483 from a small client control application from flash memory 476 (almost any small CPU can handle, such as a 100 MHz MIPS R4000 series CPU with embedded RAM) to implement network protocol stacking (also That is, the Ethernet interface) also communicates with the colocation service 210 and configures all of the devices in the client 465. The control CPU also handles the interface with the input device 469 and sends the packet back to the colocation service 210 along with the user controller data (which is subject to forward error correction protection if necessary). In addition, the control CPU 483 monitors the packet traffic (e.g., time stamped for its arrival if the packet is lost or delayed). This information is sent back to the colocation service 210 so that it can continuously monitor the network connection and adjust its delivery content accordingly. The flash memory 476 is initially loaded with a control program for controlling the CPU 483 at the time of manufacture and also loaded with a serial number that is unique to the particular client 465 unit. This sequence number allows the colocation service 210 to uniquely identify the client 465 unit.

The Bluetooth interface 484 communicates wirelessly with the input device 469 via an antenna (within the user terminal 465).

Video decompressor 486 is a low latency video decompressor configured to implement the video decompression described herein. There are many video decompression devices that are either off-the-shelf or intellectual property (IP) that can be integrated into an FPGA or a custom ASIC. The company that provides the IP of the H.264 decoder is Ocean Logic of Manly of NSW Australia. The advantage of using IP is that the compression techniques used in this article do not have to comply with the compression standard. Some standard decompressors are flexible enough to be configured to accommodate the compression techniques in this article, while some standard decompressors do not. However, with IP, there is complete flexibility in redesigning the decompressor as needed.

The output of the video decompressor is coupled to the video input subsystem 487, which couples the video to the video output of the HDMI interface 490.

The audio decompression subsystem 488 is implemented using an available standard audio decompressor, or it can be implemented as IP, or the audio decompression can be implemented in a control processor 483, which can, for example, implement Vorbis Audio Decompressor (available from Vorbis.com).

The device that implements the audio decompression is coupled to an audio output subsystem 489 that couples audio to the audio output of the HDMI interface 490.

Figure 4f shows an embodiment of the internal architecture of the client 475. As can be seen, the architecture is identical to the architecture of the client 465 except for an additional interface from a power supply adapter and an optional external DC power supply that is inserted into the wall and replaced when used. Power from the Ethernet PoE 497. The same functionality as client 465 will not be repeated below, but additional functionality is described below.

The CPU 483 communicates with the additional devices and configures the additional devices.

WiFi subsystem 482 provides wireless internet access via its antenna as an alternative to Ethernet 497. The WiFi subsystem is commercially available from a number of manufacturers including Atheros Communications of Santa Clara, CA.

USB subsystem 485 provides an alternative to Bluetooth communication for wired USB input device 479. The USB subsystem is fairly standard and easy to use in FPGAs and ASICs, and is often built into off-the-shelf devices that perform other functions such as video decompression.

The video output subsystem 487 produces a wider range of video output than the video output in the client 465. In addition to providing HDMI 490 video output, video output subsystem 487 also provides DVI-I 491, S-Video 492, and composite video 493. Moreover, when the DVI-I 491 interface is used for digital video, display capability 464 is passed back from the display device to control CPU 483, such that control CPU 483 can notify host escrow service 210 of the ability to display device 478. All of these interfaces provided by video output subsystem 487 are very standard interfaces and are readily available in many forms.

The audio output subsystem 489 outputs the audio digitally via the digital interface 494 (S/PDIF and/or Toslink) and outputs the audio in analogy via the stereo analog interface 495.

Round trip delay analysis

Of course, to achieve the benefits of the previous paragraph, the round trip delay between the user's action of using input device 421 and the result of seeing the action on display device 420 should not be greater than 70 milliseconds to 80 milliseconds. This delay must take into account all factors in the path from the input device 421 in the user premises 211 to the colocation service 210 and back to the display device 422 of the user premises 211. Figure 4b illustrates various components and networks (signals must be propagated over the network), and above these components and networks is a timeline that lists exemplary delays that may be expected in an actual implementation. Note that Figure 4b is simplified so that only the routing is shown. Other routes for the material of other features of the system are described below. A double-headed arrow (eg, arrow 453) indicates a round-trip delay, and a single-headed arrow (eg, arrow 457) indicates a one-way delay, and a "~" indicates an approximate measurement. It should be noted that there may be real-world situations where the listed delays cannot be achieved, but in many cases, in the United States, the use of DSL and cable modem connections to the user premises 211 can be described in the next section. Achieve these delays. In addition, please note that although the cellular connectivity to the Internet will certainly match the system shown, most current US cellular data systems (such as EVDO) cause extremely high latency and will not achieve what is shown in Figure 4b. The delay is shown. However, these basic principles can be implemented for future cellular technologies that may be able to implement this level of delay. In addition, there are many game and application scenarios (for example, games that do not require fast user response time, such as chess), where the delay caused by the current cellular data system (although the user can easily see it) is the game or application. The program is acceptable.

Starting from the input device 421 at the user site 211, once the user actuates the input device 421, a user control signal is sent to the client 415 (which may be a stand-alone device such as a video converter, or it may be Software or hardware executed in another device such as a PC or mobile device, and packetizing the user control signal (in UDP format in one embodiment) and giving the packet a destination address to reach the host escrow Service 210. The packet will also contain information indicating which user the control signals came from. The (these) control signal packets are then forwarded to the WAN interface 442 via a firewall/router/network address translation (NAT) device 443. The WAN interface 442 is an interface device provided to the user premises 211 by the user's Internet Service Provider (ISP). The WAN interface 442 can be a cable or DSL modem, a WiMax transceiver, a fiber optic transceiver, a cellular data interface, a power line internet protocol interface, or any other interface in many interfaces of the Internet. In addition, firewall/router/NAT device 443 (and possibly WAN interface 442) can be integrated into client 415. An example of such integration may be a mobile phone that includes software to implement the functionality of the home or office client 415 and to wirelessly route and connect to the Internet via a standard (eg, 802.11g) The building of the road.

The WAN interface 442 then routes the control signals to the interface of the "Internet Connection Point" 441, which should be referred to herein as the Internet Service Provider (ISP) of the user, to provide connectivity to the user premises 211. A facility that interfaces between the WAN and the general Internet or private network. The nature of the internet connection point will vary depending on the nature of the internet service provided. For DSL, the network connection point should normally be the headquarters of the telephone company where a DSLAM is located. For cable modems, the network connection point should typically be a cable multi-system operator (MSO) headend. For a cellular system, the network connection point should be a control room associated with the cellular tower. However, regardless of the nature of the network connection point, it then routes the control signal packets to the general Internet 410. The (these) control signal packets will then be routed to the WAN interface 441, most likely to be delivered to the colocation service 210 via a fiber optic transceiver interface. WAN 441 then routes the control signals to routing logic 409 (which can be implemented in many different ways, including Ethernet switches and routing servers), routing logic 409 evaluates the user's address and The control signals are routed to the correct server 402 for a given user.

The server 402 then uses the control signals as inputs to the game or application software being executed on the server 402 and uses the control signals to process a lower frame of the game or application. Once the next frame is generated, the video and audio are output from the server 402 to the video compressor 404. Video and audio can be output from server 402 to compressor 404 via various means. First, the compressor 404 can be built into the server 402, so compression can be implemented in the server 402 at the local end. Alternatively, the video and/or audio may be output in a packetized form via a network connection (such as an Ethernet connection) to a network that is private between the server 402 and the video compressor 404. Use the network, or a shared network (such as SAN 403). Alternatively, the video can be output via a video output connector from server 402, such as a DVI or VGA connector, and then captured by video compressor 404. In addition, the audio can be output from the server 402 as digital audio (eg, via a TOSLINK or S/PDIF connector) or as analog audio (which is digitized and encoded by the audio compression logic in the video compressor 404).

Once the video compressor 404 has retrieved the video frame and audio generated during the frame time from the server 402, the video compressor compresses the video and audio using the techniques described below. Once the video and audio are compressed, the video and audio are packetized with an address to send it back to the user's client 415 and routed to the WAN interface 441, which then goes through the general Internet. The video and audio packets are sent by the 410. The general Internet 410 then delivers the video and audio packets to the ISP network connection point 441 of the user, and the ISP network connection point 441 projects the video and audio packets. The WAN interface 442 is sent to the user premises, and the WAN interface 442 delivers the video and audio packets to the firewall/router/NAT device 443, and the firewall/router/NAT device 443 then delivers the video and audio packets to Client 415.

The client 415 decompresses the video and audio, and then displays the video on the display device 422 (or the built-in display device on the user side) and sends the audio to the display device 422 or to a separate amplifier/speaker or to the built-in display. The amplifier/speaker in the client.

In order for the user to feel that the entire process just described feels no hysteresis, the round trip delay must be less than 70 milliseconds or 80 milliseconds. Some of the delay delays in the round trip path are controlled by the colocation service 210 and/or the user, while other delay delays are not controlled by the colocation service 210 and/or the user. Nonetheless, based on analysis and testing of many real-world scenarios, the following are approximate measurements.

The one-way transmission time used to transmit the control signal 451 is typically less than one millisecond, and the round-trip routing through the user premises 452 is typically accomplished in about 1 millisecond on the Ethernet using readily available consumer-grade firewall/router/NAT switches. User ISPs vary widely in their round-trip delays 453, but for DSL and cable modem providers, we typically find round-trip delays between 10 milliseconds and 25 milliseconds. There are many variations in the round-trip delay visual traffic delivery method on the Internet 410 and whether there are any faults in the flight path (and the problems discussed below), but usually, given the distance to the destination In the case of the general Internet, the best route is provided, and the delay is mainly determined by the speed at which the light passes through the fiber. As discussed further below, we have determined 1000 为 as the approximate farthest distance that the host escrow service 210 desires from the user premises 211. At 1000 哩 (2000 哩 round trip), the actual transfer time of the signal via the Internet is approximately 22 milliseconds. The WAN interface 441 of the colocation service 210 is typically a commercial-grade fiber optic high-speed interface with negligible latency. Therefore, the general internet latency 454 is typically between 1 millisecond and 10 milliseconds. The one-way routing 455 delay via the colocation service 210 can be achieved in less than one millisecond. Server 402 will typically calculate a new frame for a game or an application for less than one frame time (16.7 milliseconds at 60 fps), so 16 milliseconds is a reasonable maximum one-way delay 456 that can be used. In one of the optimized hardware implementations of the video compression and audio compression algorithms described herein, compression 457 can be completed in 1 millisecond. In a suboptimized version, compression can take up to 6 milliseconds (of course, a more optimized version will take more time, but these implementations will affect the total latency of the round trip and require other delays to be shorter (for example, Reduce the allowable distance through the general Internet to maintain a delay of 70 milliseconds to 80 milliseconds). The round-trip delay of the Internet 454, the user ISP 453, and the user location route 452 has been considered. Therefore, the remaining consideration is the video decompression 458 delay, and the video decompression 458 is implemented in dedicated hardware or This is a software implementation on the client device 415 (such as a PC or mobile device) that varies depending on the size of the display and the performance of the decompressing CPU. Typically, decompression 458 takes between 1 millisecond and 8 milliseconds.

Thus, by adding together the worst case delays seen in practice, we can determine the worst case round trip delay that the user of the system shown in Figure 4a is expected to experience. The worst case delay is: 1+1+25+22+1+16+6+8=80ms. And, in fact, in practice (discussed below), this delay is roughly foreseen using the prototype version of the system shown in Figure 4a, using a ready-made Windows PC as a US-based device and a home DSL and cable modem connection. Round trip delay. Of course, situations that are slightly better than the worst can lead to shorter delays, but these scenarios are not reliable in developing widely used commercial services.

Resolving the delays listed in FIG. 4b on the general Internet requires that the video compressor 404 and the video decompressor 412 from FIG. 4a in the client 415 generate a packet stream with very special characteristics so that the self-host is passed. The sequence of packets generated by the entire path of the escrow service 210 to the display device 422 is not subject to delay or excessive packet loss, and in particular, is consistently suitable for the user on the Internet via the WAN interface 442 and the firewall/router/NAT 443. The bandwidth constraints available on the network connection. In addition, the video compressor must generate a stream of packets that is sufficiently robust that it can tolerate unavoidable packet loss and packet reordering that occur in normal Internet and network transmissions.

Low latency video compression

To achieve the foregoing objectives, an embodiment uses a new method for video compression that reduces the latency and peak bandwidth requirements of the transmitted video. Prior to describing such embodiments, an analysis of current video compression techniques will be provided with respect to Figures 5 and 6a-6b. Of course, if the user has sufficient bandwidth to handle the data rates required for such techniques, then such techniques can be used in accordance with the underlying principles. Please note that audio compression is not addressed in this article, except that audio compression and video compression are implemented simultaneously and synchronized with video compression. There are prior art audio compression techniques that meet the needs of this system.

5 illustrates a particular prior art for compressing video in which each individual video frame 501-503 is compressed by compression logic 520 using a particular compression algorithm to produce a series of compressed frames 511-513. One embodiment of this technique is "Motion JPEG" in which each frame is compressed according to a Joint Image Experts Group (JPEG) compression algorithm based on Discrete Cosine Transform (DCT). However, various different types of compression algorithms can be used while still adhering to these basic principles (eg, wavelet-based compression algorithms such as JPEG-2000).

One problem with this type of compression is that it reduces the data rate of each frame, but it does not exploit the similarity between successive frames, and reduces the data rate of the entire video stream. For example, as illustrated in Figure 5, assume a frame rate of 640x480x24 bits/pixel = 640*480*24/8/1024=900 kilobits/frame (KB/frame), for one For a given quality image, Motion JPEG can only compress the stream to one-tenth of the original, resulting in a stream of 90 KB/frame data. At a rate of 60 frames per second, this compression method would require a 90 KB*8 bits*60 frame/sec = 42.2 Mbps channel bandwidth, which is common to almost all home Internet connections in the United States today. It is extremely high frequency wide and is too high bandwidth for many office internet connections. In fact, if the compression method requires a constant data rate at this high frequency width and the compression method only serves one user, the compression method will consume a 100 Mbps Ethernet LAN even in an office LAN environment. A large percentage of bandwidth and a heavy burden for Ethernet switches that support LAN. Therefore, compression of motion video is inefficient compared to other compression techniques, such as the compression techniques described below. In addition, a single frame compression algorithm using lossy compression algorithms such as JPEG and JPEG-2000 produces compression artifacts that may not be apparent in still images (eg, artifacts in dense leaves in a scene may not appear as False shadows, because the eyes are not accurate, know how dense leaves appear). However, once the scene is moving, the artifacts can be highlighted because the eye detects artifacts that change between frames, despite the fact that the artifacts are in one of the scenes where the artifacts may not be apparent in the still image. This results in a sense of "background noise" in the sequence of frames, which is similar in appearance to "snow" noise that is visible during edge analog TV reception. Of course, this type of compression can still be used in the specific embodiments described herein, but in general, to avoid background noise in the scene, a given perceptual quality requires a high data rate (ie, one Low compression ratio).

Other types of compression methods (such as H.264, or Windows Media VC9, MPEG2, and MPEG4) are all more efficient when compressing a video stream because they exploit the similarities between successive frames. These technologies all rely on the same general techniques to compress video. Thus, although the H.264 standard will be described, the same general principles can be applied to various other compression algorithms. Many H.264 compressors and decompressors are available, including the x264 open source software library for compressing H.264 and the FFmpeg open source software library for decompressing H.264.

6a and 6b illustrate an exemplary prior art compression technique in which a series of uncompressed video frames 501-503, 559-561 are compressed by compression logic 620 into a series of "1 frames" 611, 671, "P maps. Frames 612-613 and "B-frames" 670. The vertical axis in Figure 6a generally represents the resulting size of each of the encoded frames (although the figures are not drawn to scale). As described above, those skilled in the art are well aware of video coding using I, B, and P frames. Briefly, I-frame 611 is a DCT-based compression of a completely uncompressed frame 501 (similar to a compressed JPEG image as described above). P-frames 612-613 are significantly smaller overall in size than I-frame 611 because the P-frame utilizes data from the previous I-frame or P-frame; that is, the P-frame contains information indicating the previous I-frame or P-frame Information on the changes between. The B-frame 670 is similar to the P-frame except that it is referred to in the frame of the frame after use of the B-frame and possibly before the frame in the reference frame.

For the following discussion, it will be assumed that the desired frame rate is 60 frames per second, each I frame is approximately 160 Kb, the average P frame and B frame are 16 Kb, and a new I frame is generated every second. Based on this set of parameters, the average data rate can be: 160 Kb + 16 Kb * 59 = 1.1 Mbps. This data rate is definitely the maximum data rate for many current broadband Internet connections to homes and offices. This technique also tends to avoid background noise problems encoded only in the frame, because the P frame and the B frame track the difference between the frames, so the compression artifacts are not easy to appear or disappear between the frames. In order to reduce the above background noise problem.

One problem with the compression method of the previous type is that although the average data rate is relatively low (for example, 1.1 Mbps), the transmission of a single I frame takes several frame times. For example, using the prior art, a 2.2 Mbps network connection (e.g., a DSL or cable modem with a 2.2 Mbps peak from the maximum available data rate 302 of Figure 3a) is typically sufficient to stream 160 Kbps at 1.1 Mbps ( Video of one I frame per 60 frames. This stream can be implemented by having the decompressor discharge the video into the queue for 1 second before decompressing the video. In 1 second, 1.1 Mb of data will be transmitted, and a 2.2 Mbps maximum available data rate will be easily adapted to this transmission, even though the available data rate may be periodically reduced by as much as 50%. Unfortunately, this prior art method resulted in a 1 second delay in video due to the 1 second video buffer at the receiver. This delay is sufficient for many prior art applications (eg, linear video playback), but is a very long delay for fast motion video games that cannot tolerate delays of more than 70 milliseconds to 80 milliseconds.

If you try to eliminate the 1 second video buffer, it still can't cause the delay of the fast action video game to be shortened enough. For example, as previously described, the use of a B-frame requires receipt of all B-frames and I-frames before an I-frame. If it is assumed that 59 non-I frames are roughly dispersed between the P frame and B, at least 29 B frames and one I frame will be received before any B frames can be displayed. Therefore, regardless of the available bandwidth of the channel, a delay of 29+1=30 frames, or a delay of 500 milliseconds, is required every sixtieth of a second. This delay is undoubtedly very long.

Therefore, another method can be to eliminate the B frame and only use the I frame and the P frame. (One of the results of this method is that for a given quality level, the data rate will increase, but for consistency in this example, continue to assume that each I frame is 160Kb in size and the average P frame size It is 16Kb, and therefore the data rate is still 1.1 Mbps). This method eliminates the inevitable delay introduced by the B-frame because the decoding of each P-frame relies only on previously received frames. The remaining problem with this approach is that the I frame is much larger than the average P frame, so that on a low frequency wide channel (as is typical in most homes and many offices), the transmission of the I frame increases the substantial delay. This is illustrated in Figure 6b. In addition to the I frame, the video stream data rate 624 is below the available maximum data rate 621, with the peak data rate 623 required for the I frame far exceeding the available maximum data rate 622 (and even the nominal maximum data rate 621). The data rate required for the P frame is less than the maximum data rate available. Even if the maximum available data rate peak of 2.2 Mbps is steadily maintained at its 2.2 Mbps peak rate, the transmission I frame will take 160 Kb/2.2 Mb = 71 ms and will drop by 50% (1.1 Mbps) if the maximum data rate 622 is available. , the transmission of the I frame will take 142 milliseconds. Therefore, the delay in transmitting the I frame falls somewhere between 71 milliseconds and 142 milliseconds. This delay is increased to the delay identified in Figure 4b. In the worst case, the delay is a total of 70 milliseconds, so this will result from the time the user actuates the input device 421 until an image appears on the display device 422. The extremely high total round trip delay from 141 milliseconds to 222 milliseconds. Also, if the maximum available data rate drops below 2.2 Mbps, the delay will increase further.

Please also note that "interfering" with an ISP at a peak data rate 623 that far exceeds the available data rate 622 typically has serious consequences. Devices in different ISPs will behave differently, but when receiving packets at a much higher data rate than the available data rate 622, the following behaviors are quite common in DSL and cable modems: (a) by placing packets in the packet Columns (introducing delays) to delay packets, (b) discarding some or all of these packets, (c) deactivating the connection for a period of time (most likely because the ISP is concerned about a malicious attack, such as a "blocking service" attack) . Therefore, transmitting a packet stream having characteristics such as the one shown in Figure 6b at a full data rate is not a viable option. The peak 623 can be queued at the colocation service 210 and sent at a data rate below the maximum available data, introducing the unacceptable delay described in the previous paragraph.

In addition, the video stream data rate sequence 624 shown in FIG. 6b is a very "moderate" video stream data rate sequence, and it is desirable for us to self-compress video from a video sequence that has little variation and has little motion. The type of data rate sequence generated (eg, is common in video conferencing where the camera is in a fixed position and moves very little, and objects in the scene (eg, the occupant is talking) exhibit little movement).

The video stream data rate sequence 634 shown in Figure 6c is what we would expect to see from video with many actions, such as video that may be produced in an animation or in a video game or in an application software. The unique sequence. Note that in addition to the I-frame peak 633, there are also P-frame peaks such as 635 and 636, which are quite large and exceed the available maximum data rate multiple times. Although the peaks of these P-frames are not as large as the I-frame peaks, they are still too large to be carried by the channel at the full data rate, and like the I-frame peaks, the peaks of these P-frames must be slowly transmitted (borrowed) This results in an increased delay).

On a high-bandwidth channel (for example, a 100 Mbps LAN, or a high-bandwidth 100 Mbps private connection), the network should be able to tolerate large peaks such as I-frame peak 633 or P-frame peak 636, and in principle, Low latency can be maintained. However, such networks are often shared by many users (for example, in an office environment), and this "peaking" data will affect the performance of the LAN, especially if the network traffic is delivered to a private shared connection. In case of (for example, from a remote data center to an office). First, keep in mind that this example is about a relatively low resolution video stream at 640x480 pixels at 60 fps. The 1920x1080 HDTV stream at 60 fps is easily handled by modern computers and monitors, and the 2560x1440 resolution display at 60 fps is increasingly available (eg, Apple, Inc.'s 30" display). To achieve a reasonable quality rating A high-motion video sequence of 1920x1080 at 60 fps may require 4.5 Mbps (using H.264 compression). If the I-frame peak is assumed to be 10 times the nominal data rate, this results in a peak of 45 Mbps and a smaller But the P-frame peaks still have to be considered. If several users are receiving video streams on the same 100 Mbps network (for example, a private network connection between an office and a data center), it is easy to see. How the peaks of video streams from several users are aligned, thereby overwhelming the bandwidth of the network and possibly overwhelming the bandwidth of the backplane of the switch that supports the user on the network. Even in a billion-bit Ethernet In the case of a road, if enough users have enough peaks to align at the same time, they can overwhelm the network or network switch. And once the 2560x1440 resolution video becomes more common, the average video stream data rate can be 9.5 Mbps, which can result in a peak data rate of 95 Mbps. Needless to say, a 100 Mbps connection between a data center and an office (which is now a particularly fast connection) can be completely from a single user. Peak traffic is overwhelmed. Therefore, even if LAN and private network connections are more tolerant of peak streaming video, streaming video with high peaks is not desirable and may require special planning by the IT department of the office. Adjustment.

Of course, for standard linear video applications, these problems are not a problem because the data rate is "smoothed" at the transmission point and the data rate of each frame is below the maximum available data rate 622, and one of the buffers in the client. Store a sequence of I, P, and B frames before decompressing the frame. Therefore, the data rate on the network remains close to the average data rate of the video stream. Unfortunately, even if the B frame is not used, this approach introduces a delay that is unacceptable for low latency applications such as video games and applications that require fast response times.

One prior art solution to reduce video streams with high peaks is to use a technique often referred to as "fixed bit rate" (CBR) encoding. Although the term CBR seems to imply that all frames are compressed to have the same bit rate (ie, size), it is generally referred to as a compression paradigm in which a certain number of frames are allowed (in the case of the present invention, one) The maximum bit rate on the frame). For example, in the case of Figure 6c, if a CBR constraint is applied to the encoding, which limits the bit rate to, for example, 70% of the nominal maximum data rate 621, the compression algorithm will limit the frames. Each of the frames is compressed such that any frame is compressed with fewer bits, which are typically compressed to use 70% greater than the nominal maximum data rate 621. As a result of this compression, it is often necessary to have more bits to maintain a "defect" position for a given quality level frame, and the image quality of such frames will be 70% less than the rated maximum data rate 621. The image quality of other frames of many bits is poor. This method can produce acceptable results for a particular type of compressed video in which (a) there is little motion or scene change expected and (b) the user can accept periodic quality degradation. A good example of a CBR-suitable application is a video conference call because there are few peaks and if the quality is temporarily degraded (for example, if the camera is shaken, resulting in significant scene motion and large peaks, there is not enough during the shaking The element can be used for high-quality image compression, which will result in degraded image quality), which is acceptable to most users. Unfortunately, CBR is not well suited for scenarios with high complexity or large amounts of motion and/or many other applications that require a fairly constant quality level.

The low latency compression logic 404 used in one embodiment uses a number of different techniques to solve a series of problems with streaming low latency compressed video while maintaining high quality. First, the low latency compression logic 404 only generates I and P frames, thereby mitigating the need to wait for a number of frame times to decode each B frame. Additionally, as illustrated in Figure 7a, in one embodiment, low latency compression logic 404 subdivides each uncompressed frame 701-760 into a series of "tiles" and individually encodes each image block. Is an I frame or a P frame. In this paper, the group of compressed I and P frames is referred to as "R frame" 711-770. In the particular example shown in Figure 7a, each uncompressed frame is subdivided into a 4 x 4 matrix of 16 image blocks. However, these basic principles are not limited to any particular subdivision scheme.

In one embodiment, the low latency compression logic 404 divides the video frame into a plurality of image blocks and encodes (ie, compresses) one image block from each frame into an I frame (ie, the image). The block is compressed as if it were a single video frame of one-sixteenth the size of the complete image, and the compression method used for this "mini" frame is the I-frame compression method, and the remaining image blocks are The encoding (ie, compression) is a P-frame (ie, the compression method used for each "mini" one-sixteenth frame is the P-frame compression method). The image blocks compressed into I frame and P frame should be referred to as "I image block" and "P image block" respectively. For each subsequent video frame, the image block to be encoded as an I image block has a change. Therefore, in a given frame time, only one of the image blocks in the video frame is an I image block, and the rest of the image blocks are P image blocks. For example, in FIG. 7a, the image block 0 of the uncompressed frame 701 is encoded into the I picture block I ₀ and the remaining picture blocks 1-15 are encoded into the P picture blocks P ₁ to P ₁₅ to generate R. Box 711. In the next uncompressed video frame 702, the image block 1 of the uncompressed frame 702 is encoded as an I image block I ₁ and the remaining image blocks 0 and 2 to 15 are encoded as P image blocks P ₀ and P ₂ to P ₁₅ to generate R frame 712. Therefore, the I image block and the P image block of the image block are gradually interleaved by time on the continuous frame. The process continues until an R image block 770 is generated and the last image block in the matrix is encoded as an I image block (i.e., I ₁₅ ). The process then begins from the beginning, producing another R frame such as frame 711 (i.e., one of the encoded image blocks 0) and the like. Although not illustrated in FIG. 7a, in one embodiment, the first R frame of the video sequence of the R frame contains only the I image block (ie, such that the subsequent P frame has reference image data for computational motion). . Alternatively, in one embodiment, the start sequence typically uses the same I picture block pattern, but it does not include P picture blocks of the picture blocks that have not been encoded with an I picture block. In other words, the particular image block is encoded without any data prior to the arrival of the first I image block, thereby avoiding the start peak in the video stream data rate 934 in Figure 9a, as will be explained further below. In addition, as described below, a variety of different sizes and shapes can be used for the image blocks while still complying with these basic principles.

The video decompression logic 412 executing on the client 415 as if each image block decompresses each image block for the same video sequence of the small I frame and the P frame, and then provides each image block to the drive display. The frame buffer of device 422. For example, image blocks 0 of the video image are decompressed and reproduced using I ₀ and P ₀ from R frames 711 through 770. Similarly, image block 1 is reconstructed using I ₁ and P ₁ from R frames 711 through 770, and the like. As described above, the decompression of the I frame and the P frame is well known in the art, and the I image block and the P image can be realized by executing a plurality of execution entities of a video decompressor in the client terminal 415. Decompression of the block. Although the process of multiplication seems to increase the computational burden on the client 415, this is not the case, since the image block itself is proportionally small relative to the number of additional processes, so the number of pixels displayed is a process and a conventional one is used. The case of the full size I frame and the P frame is the same.

This R-frame technique significantly reduces the bandwidth peaks typically associated with the I-frames illustrated in Figures 6b and 6c, as any given frame consists primarily of P-frames that are typically smaller than the I-frame. For example, assuming that a typical I frame is 160 Kb, the I image block of each of the frames illustrated in Figure 7a is approximately 1/16 or 10 Kb of this amount. Similarly, assuming a typical P frame is 16 Kb, the P frame of each of the image blocks illustrated in Figure 7a can be approximately 1 Kb. The final result is an R frame of approximately 10 Kb + 15 * 1 Kb = 25 Kb. Therefore, each 60 frame sequence will be 25Kb*60=1.5 Mbps. Therefore, in the case of 60 frames per second, this would require a channel capable of maintaining a bandwidth of 1.5 Mbps, but due to the fact that the I image block is interspersed in a 60 frame interval, the channel has a much lower peak. .

Note that in the previous example of the same hypothetical data rate for I and P frames, the average data rate was 1.1 Mbps. This is because in the previous example, only one new I frame is introduced every 60 frame time, and in this example, 16 image blocks constituting an I frame are cycled through 16 frame times, and Therefore, an equivalent of an I frame is introduced every 16 frame times, resulting in a slightly higher average data rate. In practice, although the average data rate increases, the introduction of I-frames more frequently does not linearly increase the data rate. This is due to the fact that the P-frame (or P-image block) primarily encodes the difference between the previous frame and the next frame. Therefore, if the previous frame is very similar to the next frame, the P frame will be extremely small. If the previous frame is very different from the next frame, the P frame will be extremely large. However, because the P-frame is primarily derived from the previous frame rather than derived from the actual frame, the resulting encoded frame may contain more errors than an I-frame with a sufficient number of bits (eg, visual artifacts). Moreover, in the case where one P frame follows another P frame, accumulation of errors may occur, which becomes worse in the presence of a long sequence of P frames. Now, a sophisticated video compressor will detect the fact that the quality of the image is degraded after a sequence of P frames, and if necessary, the compressor assigns more bits to subsequent P frames to boost the Quality, or replace a P-frame with an I-frame if it is the most efficient action process. Thus, when using a long P-frame sequence (eg, as in the previous example above, 59 P-frames), especially when the scene has a lot of complexity and/or motion, the P-frame follows its distance An I frame is farther and requires more bits.

Alternatively, looking at the P-frame from the opposite perspective, the P-frame following an I-frame tends to require fewer bits than the P-frame that is farther away from the I-frame. Thus, in the example shown in Figure 7a, the P-frame distance is no more than 15 frames in its previous I-frame, whereas in the previous example, a P-frame can be from an I-frame 59. Frame. Therefore, since the I frame appears more often, the P frame is smaller. Of course, the exact relative size will vary based on the nature of the video stream, but in the example of Figure 7a, if an I block is 10 Kb, the P block can average only 0.75 kb in size, resulting in 10 Kb+. 15*0.75Kb=21.25Kb, or under 60 frames per second, the data rate can be 21.25Kb*60=1.3 Mbps, or one map with one of the 1.1 Pps followed by 59 P frames. The frame stream is about 16% higher than the data rate. Again, the relative result between these two video compression methods will change the video sequence, but in general, we have empirically found that for a given quality level, using R frames requires more than using I/P frame sequences. About 20% of the bits. However, of course, the R frame significantly reduces the peak, which makes the available video sequence have a much shorter delay than the I/P frame sequence.

The R frame can be configured in a variety of different ways depending on the nature of the video sequence, the reliability of the channel, and the available data rate. In an alternate embodiment, a number of image blocks other than 16 of the 4x4 configuration are used. For example, 2 image blocks can be used in a 2x1 or 1x2 configuration, 4 image blocks can be used in a 2x2, 4x1 or 1x4 configuration, and 6 image blocks can be used in a 3x2, 2x3, 6x1 or 1x6 configuration. Medium, or 8 image blocks can be used in a 4x2 (as shown in Figure 7b), 2x4, 8x1 or 1x8 configuration. Note that these image blocks do not have to be squares, and the video frames do not have to be square or even rectangular. The image block can be decomposed into any shape that best suits the video stream and application used.

In another embodiment, the loop of the I image block and the P image block does not lock the number of image blocks. For example, in a 4x2 configuration of an 8 image block, as illustrated in Figure 7b, a 16 cycle sequence can still be used. The sequentially uncompressed frames 721, 722, 723 are each divided into 8 image blocks 0-7 and each image block is individually compressed. From R frame 731, only image block 0 is compressed into an I image block, and the remaining image blocks are compressed into P image blocks. For subsequent R frame 732, all of the 8 image blocks are compressed into P image blocks, and then for subsequent R frame 733, image block 1 is compressed into I image blocks and all other image blocks are compressed into P image blocks. Moreover, the sorting continues for 16 frames, and only one I image block is generated every other frame. Therefore, the last I image block is generated with respect to the image block 7 during the 15th frame time (in FIG. 7b). Not shown), and during the 16th frame time, R frame 780 is compressed using all P image blocks. Then, the sequence begins again by compressing the image block 0 into an I image block and compressing the other image blocks into P image blocks. As in the previous embodiment, the first frame of the entire video sequence is typically all I picture blocks, providing a reference for the subsequent P picture blocks. The loop of the I image block and the P image block does not even have to be an even multiple of the number of image blocks. For example, for 8 image blocks, each frame having an I image block may be followed by 2 frames having all P image blocks before using another I image block. In yet another embodiment, less frequent I images are required if, for example, a particular area of the known screen is more dynamic and more frequent I image blocks are needed, while other areas are more static (eg, showing a game score) Block, the specific image blocks can be sorted to have more I image blocks than other image blocks. In addition, although each frame has a single I image block as illustrated in FIGS. 7a through 7b, a plurality of I image blocks may be encoded in a single frame (depending on the bandwidth of the transmission channel). Conversely, a particular frame or frame sequence can be transmitted without an I-picture block (ie, using only P-picture blocks).

The reason for the previous paragraph's approach is that although the I image block is not scattered on each single frame and seems to cause large peaks, the behavior of the system is not so simple. Since each image block is compressed separately from other image blocks, the coding efficiency of each image block can be made low as the image block becomes smaller, because the compressor for a given image block cannot utilize other image blocks. Similar to image features and similar sports. Therefore, dividing the screen into 16 image blocks will generally result in less than one of the effective coding of the screen into 8 image blocks. However, if the screen is divided into 8 image blocks and the data of a complete I frame is introduced every 8 frames instead of every 16 frames, it results in an overall much higher data rate. Therefore, by introducing a complete I frame every 16 frames instead of every 8 frames, the total data rate is reduced. In addition, by using 8 larger image blocks instead of 16 smaller image blocks, the total data rate is reduced, which also causes the data peaks caused by the larger image blocks to be reduced to a certain extent.

In another embodiment, the low latency video compression logic 404 of Figures 7a and 7b controls the allocation of various image blocks from bit to R frames, which are set based on the known characteristics of the video sequence to be compressed. Pre-configured, or automatically based on one of the image quality in each image block. For example, in some racing video games, the front part of the player's car (which is relatively less moving in the scene) occupies most of the lower half of the screen, while the upper half of the screen is completely oncoming roads, buildings and The landscape (which is almost always in execution) fills up. If compression logic 404 assigns an equal number of bits to each image block, the image blocks (image blocks 4-7) on the lower half of the screen in uncompressed frame 721 in Figure 7b are typically compressed. It has a higher quality than the image block (image block 0-3) in the upper half of the screen in the uncompressed frame 721 in Fig. 7b. If it is known that this particular game or this particular scenario of the game has such characteristics, the cooperator 210 can configure the compression logic 404 to assign the bit ratio assigned to the image block in the top of the screen to the bottom of the screen. There are many bits in the image block. Alternatively, compression logic 404 may evaluate the compression quality of the image block after compressing the frame (using one or more of a number of compression qualities, such as peak signal to noise ratio (PNSR)), and if determined at a particular time window In the case where a particular image block consistently produces better quality results, the compression logic gradually assigns more bits to the image blocks that produce lower quality results until the various image blocks reach a similar quality level. In an alternate embodiment, compressor logic 404 allocates bits to achieve a higher quality of a particular image block or group of image blocks. For example, the logic can provide a better overall perceived appearance with a higher quality at the center of the screen than at the edges.

In one embodiment, to improve the resolution of a particular region of the video stream, the video compression logic 404 uses a smaller image block encoding than the region of the video stream with relatively little scene complexity and/or motion. Relatively more scene complexity and/or the region of the moving video stream. For example, as illustrated in Figure 8, a smaller image block is used around one of the mobile characters 805 in an area having an R frame 811 (possibly followed by a series of R frames having the same image block size ( Not shown in the figure)). Next, when the character 805 moves into a new area of the image, a smaller image block is used around the new area in the other R frame 812, as illustrated. As noted above, a variety of different sizes and shapes can be used for image blocks while still complying with these basic principles.

Although the above-described cyclic I/P image block substantially reduces the peak of the data rate of a video stream, it does not completely eliminate the peaks, especially in the case of rapidly changing or extremely complex video imaging, such as with animations. , video games and the appearance of an application software. For example, during a sudden scene transition, a complex frame may be followed by another complex frame that is completely different. Even if several I image blocks may have been ahead of the scene for only a few frame times, these image blocks are not helpful for this situation because the material of the new frame is independent of the previous I image block. In this case (and in other cases where not all things change, but most of the image changes), video compressor 404 can determine that more, if not all, of the P-picture blocks are more efficiently encoded as I-picture blocks, The result is the maximum peak in the data rate of the frame.

As discussed previously, the fact is simple. For most consumer-grade Internet connections (and many office connections), the "interference" data that exceeds the maximum data rate available at 622 and the maximum data rate of 621 shown in Figure 6c is not acceptable. Row. Note that the nominal maximum data rate of 621 (eg, "6 Mbps DSL") is essentially a sales figure for a user considering purchasing an internet connection, but it typically does not guarantee a performance level. It is not relevant for this application because one only cares about the maximum data rate 622 available when streaming video over the connection. Thus, in Figures 9a and 9c, when one describes a solution to the peak problem, the nominal maximum data rate is omitted from the figure and only the maximum data rate 922 is shown. The video stream data rate cannot exceed the maximum data rate 922 available.

To solve this problem, the first thing the video compressor 404 does is to determine a peak data rate 941, which is the data rate that the channel can easily handle. This rate can be determined by a number of techniques. One technique is to gradually transmit a test stream with a higher data rate from the colocation service 210 to the client 415 (in Figures 4a and 4b), and cause the client to report the packet loss level and Delayed feedback is provided to the host hosting service. When the packet is lost and/or the delay begins to show a sharp increase, the indication reaches the maximum available data rate 922. Thereafter, the colocation service 210 can gradually reduce the data rate of the test stream until the client 415 reports that the test stream is received at an acceptable packet loss level and the delay is near minimum for a reasonable period of time. . This determines a maximum peak data rate 941 which will then be used as one of the peak data rates for streaming video. The peak data rate 941 will fluctuate over time (eg, another user in the home begins to use the Internet connection frequently), and the client 415 will have to continuously monitor the data rate to see if the packet loss or delay is increasing, thereby indicating The available maximum data rate 922 is falling below the previously determined peak data rate 941, and if the packet is lost or the delay is increased, the maximum data rate 922 can be used to fall below the peak data rate 941. Similarly, if the client 415 finds over time that the packet loss and delay remain at an optimal level, the UE can request the video compressor to slowly increase the data rate to see if the maximum available data rate has increased (eg, in the home) Another user has stopped using the Internet connection frequently, and waits again until the packet is lost and/or the higher latency indication has exceeded the maximum data rate 922 available, and one of the peak data rates 941 can be found again. However, the rating may be higher than the level before testing an increased data rate. Thus, by using this technique (and other techniques as such), a peak data rate 941 can be found and periodically adjusted as needed. The peak data rate 941 determines the maximum data rate that can be used by the video compressor 404 to stream the video to the user. The logic for determining the peak data rate may be implemented at the user premises 211 and/or on the colocation service 210. At the user premises 211, the client device 415 performs a calculation to determine the peak data rate and transmits this information back to the colocation service 210; at the colocation service 210, one of the courier services The processor 402 performs a calculation to determine a peak data rate based on statistics received from the client 415 (eg, packet loss, delay, maximum data rate, etc.).

Figure 9a shows an example video stream data rate 934 having substantial scene complexity and/or motion that has been processed using the cyclic I/P image block compression techniques previously described and illustrated in Figures 7a, 7b, and 8. produce. The video compressor 404 has been configured to output the compressed video at an average data rate below the peak data rate 941, and note that the video stream data rate remains below the peak data rate 941 for most of the time. A comparison of the data rate 934 with the video stream data rate 634 generated using the I/P/B or I/P frame shown in Figure 6c shows that the cyclic I/P image block compression produces a smoother data rate. Also, at frame 2 times peak 952 (which is close to 2 times peak data rate 942) and frame 4 times peak 954 (which is close to 4 times peak data rate 944), the data rate exceeds peak data rate 941, which is unacceptable of. In practice, even for high-impact video from fast-changing video games, peaks above the peak data rate 941 appear in frames of less than 2%, and peaks above 2 times the peak data rate 942 rarely appear and exceed The peak of the 3 times peak data rate 943 almost never appeared. However, when such peaks occur (e.g., during a scene transition), the required data rate is necessary to produce a good quality video image.

One way to solve this problem is simply to configure the video compressor 404 such that its maximum data rate output is the peak data rate 941. Unfortunately, the resulting video output quality during these peak frames is poor because the compression algorithm is "missing" bits. The result is that compression artifacts occur when there is a sudden transition or rapid motion, and in time, the user suddenly realizes that the artifacts always appear suddenly in the presence of sudden changes or rapid movements, and the artifacts can become quite hate.

While the human visual system is quite sensitive to visual artifacts that occur during sudden changes or rapid motion, the human visual system is not particularly sensitive to detecting frame rate reductions in such situations. In fact, when these sudden changes occur, it seems that the human visual system is concentrating on tracking these changes, and it does not notice whether the frame rate temporarily drops from 60 fps to 30 fps and then immediately returns to 60 fps. Moreover, in the case of a significant conversion (such as a sudden scene change), the human visual system did not notice that the frame rate dropped to 20 fps or even 15 fps and then immediately returned to 60 fps. Since the frame rate reduction is only rare, for a human observer, it seems that the video is continuously transmitted at 60 fps.

This characteristic of the human visual system is utilized by the technique illustrated in Figure 9b. Server 402 (from Figures 4a and 4b) produces an uncompressed video output stream at a stable frame rate (60 fps in one embodiment). A timeline shows each frame 961-970 output every sixtieth of a second. Each uncompressed video frame (starting from frame 961) is output to a low latency video compressor 404, which compresses the frame in less than one frame time, thereby producing a compressed first frame. Frame 1 981. As previously described, the data generated for compressed frame 1 981 may be larger or smaller depending on a number of factors. If the data is small enough to be transmitted to the client 415 at a peak data rate 941 for one frame time (one-sixth of a second) or less, then the data is during the transmission time (xmit time) 991. Transmission (the length of the arrow indicates the duration of the transmission time). In the next frame time, the server 402 generates an uncompressed frame 2 962, compressing the uncompressed frame into a compressed frame 2 982, and the compressed frame is at a peak data rate. It is transmitted to the client 415 during a transmission time 992 that is less than one frame time.

Next, in the next frame time, the server 402 generates an uncompressed frame 3 963. After the video compressor 404 compresses the uncompressed frame, the resulting compressed frame 3 983 has more data than can be transmitted at the peak data rate 941 during a frame time. Thus, the compressed frame is transmitted during transmission time (2 times peak) 993, which occupies all of the frame time and part of the next frame time. Now, during the next frame time, the server 402 generates another uncompressed frame 4 964 and outputs it to the video compressor 404, but ignores the data and is illustrated at 974. This is because the video compressor 404 is configured to ignore other uncompressed video frames that arrive when the video compressor still transmits a previously compressed frame. Of course, the video decompressor of the client 415 will not be able to receive the frame 4, but the video decompressor only continues to display the frame 3 on the display device 422 for 2 frame times (ie, temporarily the frame rate is self-contained). 60 fps down to 30 fps).

For the next block 5, the server 402 outputs uncompressed frame 5 965, compressing the uncompressed frame into compressed frame 5 985 and transmitting it in a frame during transmission time 995 The compressed frame. The video decompressor of client 415 decompresses frame 5 and displays it on display device 422. Next, the server 402 outputs the uncompressed frame 6 966, and the video compressor 404 compresses the uncompressed frame into the compressed frame 6 986, but at this time, the resulting data is extremely large. The compressed frame is transmitted during the transmission time (4 times peak) 996 at the peak data rate 941, but it takes almost 4 frame times to transmit the frame. During the next three frame times, the video compressor 404 ignores the three frames from the server 402, and the decompressor of the client 415 stably holds the frame 6 on the display device 422 for up to four frames. Time (ie, temporarily reduce the frame rate from 60 fps to 15 fps). Finally, server 402 outputs block 10 970, which compresses the frame into compressed frame 10 987, and the compressed frame is transmitted during transmission time 997, and the client 415 The decompressor decompresses frame 10 and displays it on display device 422, and again, the video resumes at 60 fps.

Please note that although the video compressor 404 discards the video frame from the video stream generated by the server 402, the video compressor does not discard the audio data, regardless of the form in which the audio system is transmitted, the video compressor is When the video frame is discarded, the audio data is continuously compressed and the audio data is transmitted to the client 415, and the user continues to decompress the audio data and provide the audio to any device used by the user to play the audio. Therefore, the audio remains undiminished during the period in which the frame is discarded. The compressed audio consumes a relatively small percentage of the bandwidth compared to the compressed video, and thus has no significant impact on the overall data rate. Although this is not illustrated in any of the data rate graphs, there is always a data rate capacity reserved for the compressed audio stream in the peak data rate 941.

The example just described in Figure 9b is chosen to illustrate how the frame rate drops during the peak data rate, but it is not stated that even when using the cyclic I/P image block technique described earlier, even at high scene complexity These high data rate peaks and consequent discarding frames are rare during periods of high activity (such as in video games, animations, and sequences appearing in an application software). Therefore, the reduced frame rate rarely occurs and is temporary, and the human visual system does not detect a reduced frame rate.

If the frame rate reduction mechanism just described is applied to the video stream data rate illustrated in Figure 9a, the resulting video stream data rate is illustrated in Figure 9c. In this example, the 2x peak 952 has been reduced to a flattened 2x peak 953, and the 4x peak 955 has been reduced to a flattened 4x peak 955, and the entire video stream data rate 934 remains at the peak data rate 941. Or below.

Thus, using the techniques described above, a high-motion video stream can be transmitted over a general Internet and via a consumer-grade Internet connection with low latency. In addition, in an office environment, on a LAN (for example, 100 Mbs Ethernet or 802.11g wireless network) or on a private network (for example, between a data center and an office) On a Mbps connection, a high-motion video stream with no peaks can be transmitted, allowing multiple users (for example, 1920x1080 to transmit at 60 fps at 4.5 Mbps) to use the LAN or shared private data connection without overlapping peaks. , where overlapping peaks will overwhelm the network or network switch backplane.

Data rate adjustment

In one embodiment, the colocation service 210 initially evaluates the channel's available maximum data rate 622 and latency to determine the appropriate data rate for the video stream and dynamically adjusts the data rate in response to the response. To adjust the data rate, the colocation service 210 can, for example, modify the image resolution and/or the number of frames per second of the video stream to be sent to the client 415. In addition, the colocation service can adjust the quality level of the compressed image. When the resolution of the video stream is changed (e.g., from 1280 x 720 resolution to 640 x 360), the video decompression logic 412 on the client 415 can scale the image to maintain the same image size on the display. .

In an embodiment, the colocation service 210 suspends the game in the event that the channel is completely abandoned. In the case of a multiplayer game, the colocation service reports to other users that the user has abandoned the game and/or suspended the game of other users.

Discarded or delayed packet

In an embodiment, if the packet between the video compressor 404 and the client 415 in FIG. 4a or 4b is lost, or because the arrival is too late, the packet cannot be decompressed and the delay requirement of the decompressed frame is satisfied. The packets are received out of order, and the lost data, the video decompression logic 412 can reduce visual artifacts. In a streaming I/P frame implementation, if there is a lost/delayed packet, the entire screen is affected, which may completely freeze the screen for a period of time or display other widescreen visual artifacts. For example, if a lost/delayed packet results in the loss of an I frame, the decompressor will refer to one of all subsequent P frames until a new I frame is received. If a P frame is lost, it will affect the subsequent P frame of the entire screen. Depending on the time elapsed before the appearance of an I frame, this loss will have a longer or shorter visual impact. Using the interleaved I/P image blocks as shown in Figures 7a and 7b, a lost/delayed packet is less likely to affect the entire screen because it only affects the image blocks contained in the affected packet. If the data of each image block is sent in a separate packet, if one packet is lost, it will only affect one image block. Of course, the duration of the visual artifact will depend on whether or not an I image block packet is lost; and if a P image block is lost, how many frames will be consumed before an I image block appears. However, if the I image frame is updated frequently (perhaps with each frame), the other image blocks will be unaffected even if one image block on the screen is affected. In addition, if an event causes several packets to be lost at the same time (for example, one of the DSL lines temporarily interrupts the power supply peak of the data stream), some of the image blocks are more affected than other image blocks, but because Some image blocks can be quickly renewed with a new I image block, so these image blocks will only be temporarily affected. In addition, for a stream I/P frame implementation, the I frames are not only the most critical frames, but the I frames are also very large. Therefore, if there is an event that causes a discard/delay packet, then The probability that an I frame is affected will be larger than a much smaller I image block (ie, if any part of an I frame is lost, it will not be possible to decompress the I frame). For all of these reasons, the use of I/P image blocks results in much less visual artifacts than I/P frames when packets are dropped/delayed.

An embodiment attempts to reduce the impact of lost packets by intelligently encapsulating compressed video blocks in Transmission Control Protocol (TCP) packets or User Datagram Protocol (UDP) packets. For example, in one embodiment, the image block is aligned with the packet boundary as much as possible. Figure 10a illustrates the manner in which image blocks can be packaged in a series of packets 1001-1005 without implementing this feature. In particular, in Figure 10a, the image blocks span the packet boundaries and are packaged inefficiently such that the loss of a single packet results in the loss of multiple frames. For example, if the packet 1003 or 1004 is lost, the three image blocks will be lost, resulting in visual artifacts.

In contrast, Figure 10b illustrates image block packaging logic 1010 for intelligently packaging image blocks in a packet to reduce the effects of packet loss. First, image block wrapper logic 1010 aligns the image block with the packet boundaries. Therefore, the image blocks T1, T3, T4, T7, and T2 are aligned with the boundaries of the packets 1001-1005, respectively. The image block wrapper logic also attempts to fit the image block into the packet in the most efficient way possible, without crossing the packet boundary. Based on the size of each of the image blocks, the image blocks T1 and T6 are combined in one packet 1001; the image blocks T3 and T5 are combined in one packet 1002; and the image blocks T4 and T8 are combined in one packet 1003. The image block T7 is added to the packet 1004; and the image block T2 is added to the packet 1005. Therefore, according to this scheme, a single packet loss will result in the loss of no more than 2 image blocks (rather than the 3 image blocks as illustrated in Figure 10a).

An additional benefit of one of the embodiments shown in Figure 10b is that the image blocks are transmitted in an order different from the order in which they are displayed in the image. In this way, if adjacent packets are lost due to the same event that interferes with the transmission, they will affect areas that are not close to each other on the screen, thereby creating less noticeable artifacts on the display.

An embodiment uses Forward Error Correction (FEC) techniques to protect a particular portion of a video stream from channel errors. As is well known in the art, FEC techniques such as Reed-Solomon and Viterbi generate error correction data information and attach it to data transmitted via a communication channel. If an error occurs in the basic data (for example, an I frame), the FEC can be used to correct the error.

The FEC code increases the data rate of the transmission, so ideally, the FEC codes are used only when they are most needed. If you are sending data that does not cause a significant visual artifact, it may be better to protect the data without using an FEC code. For example, a P-image block immediately before a lost I-picture block will only produce a one-hundredth of a second visual artifact on the screen (ie, on an unupdated image block on the screen). ). This visual artifact is hardly detectable by the human eye. Since a plurality of P image blocks are far behind an I image block, the loss of a P image block becomes more and more obvious. For example, if an image block cycle pattern is an I image block of 15 P image blocks before an I image block can be obtained again, if the P image block immediately after an I image block is lost, This will cause the image block to display an erroneous image for 15 frames (which will be 250 milliseconds at 60 fps). The human eye will easily detect a 250 millisecond interrupt in a stream. Therefore, the farther a P image block is from a new I image block (that is, the closer a P image block is to an I image block), the more obvious the artifact is. As discussed earlier, in general, the closer a P-picture block is to an I-picture block, the smaller the data of the P-picture block. Therefore, the P image block after the I image block is not only more important in preventing loss, but also the size of the P image block is small. Moreover, in general, the smaller the data that needs to be protected, the smaller the FEC code required to protect the data.

Thus, as illustrated in Figure 11a, in one embodiment, due to the importance of the I-picture block in the video stream, only the I-picture block has the FEC code. Therefore, FEC 1101 contains an error correction code for I picture block 1100, and FEC 1104 contains an error correction code for I picture block 1103. In this embodiment, the FEC of the P-picture block is not generated.

In one embodiment illustrated in Figure 11b, an FEC code of a P-Video block that is likely to cause visual artifacts if lost is also generated. In this embodiment, the FEC 1105 provides an error correction code for the first 3 P blocks, but does not provide an error correction code for the subsequent P block. In another embodiment, the FEC code of the P image block with the smallest data size is generated (which will tend to self-select the P image block that appears fastest after an I image block, and the P image blocks are more important in terms of protection) .

In another embodiment, instead of transmitting an FEC code and an image block, the image block is transmitted twice, each time in a different packet. If one packet is lost/delayed, another packet is used.

In one embodiment shown in FIG. 11c, FEC codes 1111 and 1113 of audio packets 1110 and 1112 are generated, respectively, which are simultaneously transmitted from the colocation service with the video. Maintaining the integrity of the audio in a video stream is particularly important because distorted audio (eg, clicks or clicks) can result in a particularly unpleasant user experience. These FEC codes help ensure that the audio content is reproduced at the client computer 415 without distortion.

In another embodiment, instead of transmitting an FEC code and an audio material, the audio data is transmitted twice, each time in a different packet. If one packet is lost/delayed, another packet is used.

In addition, in one embodiment illustrated in FIG. 11d, FEC codes 1121 and 1123 are used for user input commands 11.20 and 1122 (eg, button presses) that are uploaded from the client 415 to the colocation service 210, respectively. This is very important because missing a video game or a button press or a mouse movement in an application can lead to an unpleasant user experience.

In another embodiment, instead of transmitting an FEC code and user input command data, the user input command data is transmitted twice, each time in a different packet. If one packet is lost/delayed, another packet is used.

In one embodiment, the colocation service 210 evaluates the quality of the communication channel with the client 415 to determine whether to use FEC, and if FEC is used, determines which portions of the video, audio, and user commands should be applied to the FEC. The "quality" of the rating channel may include functions such as evaluating packet loss, delay, etc., as described above. If the channel is particularly unreliable, the colocation service 210 can apply FEC to all I block, P block, audio, and user commands. Conversely, if the channel is reliable, the colocation service 210 may only apply the FEC to the audio and user commands, or may not apply the FEC to the audio or video, or may not use the FEC at all. Various other arrangements of FEC applications can be used while still complying with these basic principles. In an embodiment, the colocation service 210 continues to monitor the status of the channel and change the FEC policy accordingly.

In another embodiment, referring to Figures 4a and 4b, when a packet is lost/delayed, resulting in loss of image block data, or if (possibly) due to a particularly unfavorable packet loss, FEC cannot correct the lost image block data. The client 415 evaluates how many frames remain before receiving a new I image block and compares its round trip delay from the user terminal 415 to the colocation service 210. If the round trip delay is less than the number of frames before a new I frame that will arrive, the client 415 sends a message requesting a new I block to the colocation service 210. The message is sent to the video compressor 404 and is not a P image block of the image block in which the data has been lost. The video compressor generates an I image block. If the system shown in Figures 4a and 4b is designed to provide a round-trip delay of typically less than 80 milliseconds, this will cause an image block to be corrected within 80 milliseconds (at 60 fps, the frame has a duration of 16.67 milliseconds). Time, therefore, in the full frame time, the 80 millisecond delay will result in a corrected image block in 83.33 milliseconds, 83.33 milliseconds for 5 frame times, which is a significant interruption, but far less than (for example) 15 The 250 ms interrupt of the frame is obvious). When the compressor 404 does not generate the I image block in its normal cyclic order, if the I image block causes the bandwidth of the frame to exceed the available bandwidth, the compressor 404 will delay the loop of other image blocks to make other image blocks. Receiving a P-picture block during the frame time (even if one image block would normally be expected to be an I-picture block during the frame), and then starting from the next frame, typically the loop will continue and the previous picture is typically received The image block of one of the I image blocks in the frame will receive an I image block. Although this action temporarily delays the phase of the R frame cycle, it is usually not visually apparent.

Video and audio compressor/decompressor implementation

Figure 12 illustrates a particular embodiment in which eight image blocks are compressed in parallel using a multi-core and/or multi-processor 1200. In one embodiment, a dual processor, quad core Xeon CPU computer system executing at 2.66 GHz or higher is used, and each core implements the open source code x264 H.264 compressor as a separate handler. However, a variety of other hardware/software configurations can be used while still complying with these basic principles. For example, each of the CPU cores can be replaced with an H.264 compressor implemented in an FPGA. In the example shown in FIG. 12, the core 1201-1208 is used to process the I image block and the P image block simultaneously as eight independent threads. As is well known in the art, current multi-core and multi-processor computer systems are inherently capable of being integrated with multi-threaded operating systems such as the Microsoft Windows XP Professional Edition (64-bit or 32-bit version) and Linux. Multi-line processing.

In the embodiment illustrated in FIG. 12, since each of the eight cores is responsible for only one image block, each core is substantially independent of other core operations, and each core executes one of x264 to execute the individual individually. A DVI capture card based on the Fast Peripheral Component Interconnect x1 (PCI Express x1) (such as the Sendero Video Imaging IP Development Board from Microtronix from Oosterhout, The Netherlands) for 640x480, 800x600 Or 1280x720 resolution captures uncompressed video, and the FPGA on the card uses Direct Memory Access (DMA) to transfer the captured video to system RAM via the DVI bus. The image blocks are arranged in a 4x2 configuration 1205 (although the image blocks are illustrated as square image blocks, in this embodiment, the image blocks have a resolution of 160x240). Each execution individual of x264 is configured to compress one of the eight 160x240 image blocks and synchronize the image blocks such that after the initial I image block is compressed, each core enters a cycle, each frame It is out of phase with another frame to compress one I block, followed by seven P blocks, and is illustrated in FIG.

In each frame time, the resulting compressed image blocks are combined into a packet stream using the techniques previously described, and the compressed image blocks are then transmitted to a destination client 415.

Although not illustrated in FIG. 12, if the data rate of the combined 8 image blocks exceeds a specified peak data rate 941, it is necessary to temporarily suspend all 8 x264 processing programs for a plurality of frame times until the combined 8 images The data of the block has been transferred.

In one embodiment, the client 415 is implemented as a software that executes FFmpeg's eight execution entities on the PC. A receiving process receives the eight image blocks and routes each image block to an FFmpeg executing entity, which decompresses the image block and renders the image block on a suitable image block location on display device 422.

The client 415 receives keyboard, mouse or game controller inputs from the input device driver of the PC and transmits them to the server 402. The server 402 then applies the received input device data and applies the data to a game or application executing on the server 402, which is a PC that executes Windows using an Intel 2.16 GHz Core Duo CPU. Server 402 then generates a new frame and outputs the new frame via its DVI output from the motherboard based graphics system or via the DVI output of the NVIDIA 8800 GTX Fast Peripheral Component Interconnect (PCI Express) card.

At the same time, the server 402 outputs the audio generated by the game or the application via its digital audio output (for example, S/PDIF), and the digital audio output is coupled to the digit on the dual quad-core Xeon-based PC that implements video compression. Audio input. A Vorbis open source coded audio compressor is used to compress audio in synchronism with any video available to the processor thread. In one embodiment, the core that completes compression of its image block first performs audio compression. The compressed audio is then transmitted with the compressed video and the compressed audio is decompressed at the client 415 using a Vorbis audio decompressor.

Hosting service server center distribution

Light travels in glass (such as an optical fiber) at a fraction of the speed of light in a vacuum, and thus, the exact rate of propagation of light in the fiber can be determined. However, in practice, considering the timing of routing delays, transmission inefficiencies, and other additional burdens, we have observed that the optimal delay on the Internet reflects a transmission speed that is closer to 50% of the speed of light. Thus, an optimal 1000 哩 round trip delay is approximately 22 milliseconds and an optimal 3000 哩 round trip delay is approximately 64 milliseconds. Therefore, a single server on a US coast may be too far away to provide the required delay to a server on another coast (which may be outside 3000 miles). However, as illustrated in Figure 13a, if the server center 1300 of the colocation service 210 is located at the center of the United States (e.g., Kansas, Nebraska, etc.) such that the distance to any point in the continental United States is approximately 1500 ft or greater. Less, the round-trip internet latency can be as low as 32 milliseconds. Referring to Figure 4b, please note that although the worst case delay allowed by the user ISP 453 is 25 milliseconds, in general, we have observed delays in the DSL and cable modem systems that are closer to 10 milliseconds to 15 milliseconds. In addition, FIG. 4b assumes that the maximum distance from the user premises 211 to the colocation service 210 is 1000 哩. Thus, the input device 421 is actuated from a user and on the display device 422 with a typical user ISP round trip delay of 15 milliseconds and a maximum internet distance of 1500 哩 (with a round trip delay of 32 milliseconds). The total round trip delay for seeing a response is 1+1+15+32+1+16+6+8=80 milliseconds. Therefore, the response time of 80 milliseconds can usually be achieved via a network distance of 1500 。. This would allow any user site in the continental United States with a sufficiently short user ISP delay 453 to access a single server center located in the center.

In another embodiment illustrated in Figure 13b, the colocation service 210 server centers HS1-HS6 are strategically located around the United States (or other geographic area), with some of the larger colocation service server centers Close to population-intensive centers (eg, HS2 and HS5). In one embodiment, server centers HS1-HS6 exchange information via network 1301, which may be the Internet or a private network or a combination of both. With multiple server centers, the service can be provided to users with high user ISP latency 453 at a lower latency.

While the distance on the Internet is definitely a factor that affects the round-trip delay through the Internet, sometimes other factors that are essentially unrelated to latency begin to work. Sometimes, a packet stream is sent over the Internet to a remote location and sent back, causing delays caused by long loops. Sometimes, there are routing devices on the path that are not operating properly, resulting in a delay in transmission. Sometimes, there is a traffic that overloads a path, which introduces a delay. And sometimes, there is a failure to make the user's ISP completely unable to route to a given destination. Thus, while the general Internet typically provides a connection from one point to another with a fairly reliable and optimal route and a delay determined primarily by distance (especially long distance connections that result in routes outside the user's local area) However, this reliability and delay from a user's location to a given destination on a general Internet is never guaranteed, and this reliability and delay are often not achieved.

In one embodiment, when the user client 415 is initially connected to the colocation service 210 to play a video game or use an application, the client (eg, using the techniques described above) is hosted with the available host after startup. Each of the service server centers HS1-HS6 communicates. If the delay of a particular connection is low enough, the connection is used. In one embodiment, the client communicates with all or a subset of the hosted service server centers and selects a hosted service server center with the lowest latency connection. The client may select a service center with the lowest latency connection, or the service center may identify the service center with the lowest latency connection and provide this information (eg, in the form of an internet address) to the client.

If a particular hosted service server center is overloaded and/or the user's game or application can tolerate the delay of another hosted service server center with less load, the client 415 can be redirected to another A host hosting service server center. In this case, the game or application being executed by the user will be suspended on the server 402 at the user's overloaded server center, and the game or application status data can be transferred to another colocation service server. Server 402 at the center. Then you can restart the game or the app. In one embodiment, the colocation service 210 will wait until the game or application has reached a natural pause point (eg, between levels in a game, or "storage" in the user's originating application. This transfer is only performed after the operation. In yet another embodiment, the colocation service 210 will wait until the user activity ceases for a specified period of time (eg, 1 minute) and then the transfer can be initiated at that time.

As described above, in one embodiment, the colocation service 210 subscribes to an internet bypass service 440 of FIG. 14 to attempt to provide a guaranteed delay to its client. As used herein, Internet bypass service is a service that provides private network routes with guaranteed features (eg, latency, data rate, etc.) from one point to another on the Internet. For example, if the colocation service 210 receives a large amount of traffic from the user using the AT&T DSL service provided by San Francisco, instead of delivering to the San Francisco General Office of AT&T, the colocation service 210 can be provided from one service. The person (possibly AT&T itself or another provider) leases a high-capacity private data connection between one or more of the server centers of the San Francisco General Office and the colocation service 210. Then, if the route from all the hosted service server centers HS1-HS6 via the general Internet to one of San Francisco's users using AT&T DSL causes excessive delay, the private data connection can be used instead. Although the private data connection is usually more expensive than the route through the general Internet, as long as the route accounts for a small percentage of the user's colocation service 210 connection, the total cost impact will be low and the user Can experience a more stable service experience.

In the event of a power failure, the server center often has two sets of backup power supplies. The first group is typically a backup power source from a battery (or from an alternate ready-to-use energy source, such as a flywheel that remains operational and attached to a generator) that provides power and maintains the server immediately in the event of a power failure The center performs. If the power failure is temporary and the power line is quickly restored (for example, within one minute), the battery is sufficient to maintain the server center. However, if the power failure continues for a longer period of time, the generator (eg, a diesel generator) is typically activated, which replaces the battery and can operate with fuel. These generators are very expensive because they must be able to generate as much power as the power that the server center typically receives from the power line.

In one embodiment, each of the colocation services HS1-HS5 shares user data with each other such that if a server center experiences a power failure, it can suspend the ongoing game and application, and then The game or application status data is transmitted from each server 402 to the server 402 at the other server center, and then the user 415, which notifies each user, directs its communication to the new server 402. If such a situation rarely occurs, it may be acceptable to transfer a user to one of the hosted service server centers that does not provide the optimal delay (ie, the user only needs to be in the duration of the power failure) Tolerate higher latency), this will allow for a wider range of options for transferring users. For example, since the time zone is different in the United States, users on the East Coast may sleep at 11:30 PM, while users on the West Coast start using the peak of video games at 8:30 PM. If one of the West Coast hosted service server centers experiences a power failure at that time, there may not be enough other hosts to host the West Coast Server 402 at the Service Server Center to handle all users. In this case, some of the users can be transferred to the colocation server server with the available server 402 on the East Coast, and the only impact on the user will be a higher latency. Once the user has transferred from the server center that has been powered off, the server center can then begin to shut down its servers and devices in an orderly manner to shut down all devices before the battery (or other alternate ready-to-use power) is exhausted. . In this way, the cost of the generator at the server center can be avoided.

In an embodiment, during heavy load times of the colocation service 210 (due to peak user load, or because one or more server center failures), the user is based on the game or application they are using. Delay requests are transferred to other server centers. Therefore, in the case of limited supply, users who use games or applications that require low latency will prefer the available low latency server connections.

Hosting service feature

Figure 15 illustrates one embodiment of a component of a server hub of one of the colocation services 210 used in the following feature description. As with the colocation service 210 illustrated in FIG. 2a, such components of the server center are controlled and coordinated by the colocation service 210 control system 401 unless otherwise defined.

The incoming Internet traffic 1501 from the user client 415 is directed to the incoming route 1502. Typically, incoming Internet traffic 1501 will enter the server center via a high-speed fiber-optic connection to the Internet, but any network connectivity component with sufficient bandwidth, reliability, and low latency will meet this requirement. The incoming route 1502 is a system of a network (which may be implemented as an Ethernet network, a fiber channel network, or via any other transport component) and a routing server supporting the switches, the system Get the packets and send each packet to the appropriate app/game ("app/game") server 1521-1525. In one embodiment, a packet passed to a particular application/game server represents a subset of the data received from the client and/or may be other components in the data center (eg, such as gateways and routers) Network connection component) translation/change. In some cases, for example, if a game or application is executing concurrently on multiple servers, the packets are delivered to more than one server 1521-1525 at a time. The RAID arrays 1511-1512 are connected to the incoming routing network 1502 such that the application/game server 1521-1525 can read and write to the RAID arrays 1511-1512. In addition, RAID array 1515 (which may be implemented as multiple RAID arrays) is also coupled to incoming routing 1502 and may read material from RAID array 1515 from application/game servers 1521-1525. Inbound routing 1502 can be implemented in a number of prior art network architectures, including a tree structure of switches, wherein the incoming Internet traffic 1501 is located at the root of the tree; implemented in interconnecting all of the various devices In a mesh structure; or as a series of interconnected subnetworks, where centralized communication between devices communicating with each other is concentrated with other devices. One type of network configuration is a SAN, although it is typically used for storage devices, but it can also be used for general high-speed data transfer between devices. In addition, the application/game servers 1521-1525 can each have multiple network connections to the incoming route 1502. For example, the servers 1521-1525 can have one network connection to one of the sub-networks attached to the RAID arrays 1511-1512 and another network connection to a sub-network attached to other devices.

As previously described with respect to server 402 in the embodiment illustrated in Figure 4a, application/game servers 1521-1525 may be configured as identical, slightly different or completely different servers. In one embodiment, each user typically uses at least one application/game server 1521-1525 when using the colocation service. For the sake of simplicity, we should assume that a given user is using the application/game server 1521, but one user can use multiple servers and multiple users can share a single application/game server. 1521-1525. The user's control input (sent from client 415 as previously described) is received as incoming Internet traffic 1501 and routed to application/game server 1521 via incoming route 1502. The application/game server 1521 uses the user's control input as a control input for the game or application executing on the server, and calculates the next frame of the video and the audio associated with the frame. The application/game server 1521 then outputs the uncompressed video/audio 1529 to the shared video compression 1530. The application/game server can output uncompressed video via any component (including one or more billion-bit Ethernet connections), but in one embodiment, the video is output via a DVI connection and the audio is output. And other compression and communication channel status information is output via a universal serial bus (USB) connection.

The shared video compression 1530 compresses uncompressed video and audio from the application/game server 1521-1525. This compression can be performed entirely in the hardware or in the hardware in which the software is executed. There may be a dedicated compressor for each application/game server 1521-1525, or, if the compressors are fast enough, can be compressed from more than one application/game server 1521-1525 using a given compressor. Video/audio. For example, at 60 fps, a video frame time is 16.67 milliseconds. If a compressor can compress a frame within 1 millisecond, the compressor can be used to compress video from up to 16 application/game servers 1521-1525 by taking input from the server one by one. Audio, wherein the compressor stores the state of each video/audio compression process and switches the context of the context as the compressor cycles between video streams/audio streams from the servers. This makes it possible to greatly reduce the cost of compressing hardware. Since the different servers will complete the frame at different times, in one embodiment, the compressor resources are in a shared storage component (eg, RAM, flash memory) having a state for storing each compression process. In the shared pool 1530, and when the server 1521-1525 frame is completed and ready to be compressed, a control component determines which compression resource is available at that time, provides the state of the compression process of the server for the compressed resource and is to be compressed. Uncompressed video/audio frame.

Please note that one of the states of the compression process of each server includes information about the compression itself (such as the buffered data of the decompressed frame of the previous frame, which can be used as a reference for the P image block), and the video output. Resolution, quality of compression, image block structure, bit allocation per image block, compression quality, audio format (eg stereo, surround, Dolby) AC-3). However, the compression process state also includes communication channel status information regarding the peak data rate 941 and whether a previous frame is currently being output (as illustrated in Figure 9b) (and as a result, the current frame should be ignored), and possibly, Whether there are channel characteristics (such as excessive packet loss) that should be considered when compressing, these channel characteristics can affect the decision about compression (for example, in terms of the frequency of use of I image blocks, etc.). Since the peak data rate 941 or other channel characteristics change over time, as determined by the application/game server 1521-1525 that supports each user monitoring the data sent from the client 415, the application/game server 1521- The 1525 sends the relevant information to the shared hardware compression 1530.

The shared hardware compression 1530 also encapsulates the compressed video/audio using components such as those previously described, and where appropriate, applies FEC codes, duplicates specific data, or takes other steps to fully ensure video/audio data. The streaming can be received and decompressed by the client 415 with the highest possible quality and reliability.

Some applications, such as the applications described below, require that the video/audio output of a given application/game server 1521-1525 be simultaneously available in multiple resolutions (or in a variety of other formats). If the application/game server 1521-1525 thus notifies the shared hardware compression 1530 resource, the application/game server 1521-1525 will be simultaneously compressed in different formats, different resolutions, and/or in different packet/error correction structures. Uncompressed video audio 1529. In some cases, some compression resources may be shared between multiple compression processes that compress the same video/audio (eg, in many compression algorithms, there is a step of scaling the image to multiple sizes before compression. If you need to output images of different sizes, you can use this step to simultaneously servo several compression processes). In other cases, each format requires a separate compression of resources. In any case, the compressed video/audio 1539 of all the various resolutions and formats required for a given application/game server 1521-1525 (which can be a server or a number of servers) is simultaneously output to Outbound route 1540. In one embodiment, the output of the compressed video/audio 1539 is in UDP format, so the output is a one-way stream of packets.

The outgoing routing network 1540 includes a series of routing servers and switches that will each pass through the outgoing Internet traffic 1599 interface (which is typically connected to one of the optical interfaces of the Internet). The compressed video/audio stream is directed to the desired user or other destination(s), and/or sent back to the delay buffer 1515, and/or sent back to the incoming route 1502, and/or via a private Use the network (not shown) for output for video distribution. Note that (as described below), the outgoing route 1540 can simultaneously output a given video/audio stream to multiple destinations. In one embodiment, this output is implemented using Internet Protocol (IP) multicast, where a broadcast is intended to simultaneously stream to a given UDP stream of multiple destinations, and the broadcast is by the outgoing route 1540. The routing server and switch are duplicated. The plurality of destinations of the broadcast may be via the Internet to the user terminals 415 of the plurality of users, via the routing route 1502 to the plurality of application/game servers 1521-1525 and/or to one or more delay buffers 1515. . Thus, the output of a given server 1521-1522 is compressed into one or more formats and each compressed stream is directed to one or more destinations.

Moreover, in another embodiment, if multiple application/game servers 1521-1525 are used simultaneously by one user (eg, in a parallel processing configuration for generating a 3D output of a complex scene) and each A server generates a portion of the resulting image, and the video outputs of the plurality of servers 1521-1525 can be combined into a combined frame by the shared hardware compression 1530, and thereafter, the combined frame is disposed as described above, as if The image is from a single application/game server 1521-1525.

Note that in one embodiment, a copy of all of the video generated by the application/game server 1521-1525 (at least the resolution of the video viewed by the user or higher resolution) is recorded in the delay buffer. At least a certain number of minutes in 1515 (in one embodiment, 15 minutes). This record allows each user to "reverse" the video from each session to review previous work or merits (in the case of a game). Thus, in one embodiment, each compressed video/audio output 1539 being forwarded to a user client 415 can also be multicast to the delay buffer 1515. After storing the video/audio on the delay buffer 1515, a directory on the delay buffer 1515 provides the network address and delay of the application/game server 1521-1525, which is the source of the delayed video/audio. A cross-reference between the location on buffer 1515 (where delayed video/audio can be found).

Live, instantly watchable, playable game

The application/game server 1521-1525 can be used not only to execute a given application or video game for the user, but also to generate a user interface application for the colocation service 210, which supports generation via the host Guide to the service 210 and other features. In Figure 16, a screen capture of one of the user interface applications, a "Game Finder" screen is shown. This particular user interface screen allows one user to watch 15 games played (or delayed) by other users on the spot. Each of the "shrink" video windows (such as 1600) is a live video video in motion that displays video from a user's game. The view shown in the thumbnail can be the same view seen by the user, or the view can be a delayed view (for example, if a user is playing a battle game, a user may not want other users to see it) Where, the user can choose to delay any view of his gameplay for a period of time, ie 10 minutes). The view can also be a camera view of a game that is different from the view of any user. Through menu selection (not shown in this illustration), a user can select a choice to watch the game immediately based on various criteria. As a small sample of exemplary choices, the user may choose to randomly select a game (such as the game shown in Figure 16), all games of the same type (played by different players), top players of only one game, games A player of a given level, or a lower level player (for example, the player is learning the basic rules), a player who is a "partner" (or competitor), a game with the most audience, and the like.

Please note that usually, each user will decide whether the video from their game or application is available for viewing by other users, and if so, which other users and other users can see when they can watch, and can only delay viewing. .

The application/game server 1521-1525 that produces the user interface screen shown in FIG. 16 acquires 15 videos by transmitting a message to the application/game server 1521-1525 for each user/ An audio feed that requests video/audio from each of the user's games. The message is sent via the incoming route 1502 or another network. The message may include the size and format of the requested video/audio and will identify the user viewing the user interface screen. A given user may select the "private" mode and do not permit any other user to view the video/audio of the game (from the user's viewing point or from another viewing point), or as in the previous paragraph Description, a user may choose to allow viewing of video/audio from their game, but the video/audio to be viewed is delayed. The user application/game server 1521-1525 receiving and accepting a request to allow the user's video/audio to be viewed will respond to the request server as such, and the server will also notify the shared hardware compression 1530. The need to generate an additional compressed video stream in the requested format or screen size (assuming the format and screen size are different from the format and screen size that has been generated), and the server will also indicate the purpose of the compressed video. Ground (ie, the request server). If only the requested video/audio is delayed, the requesting application/game server 1521-1525 will be notified as such, and the server will look up the location of the video/audio in the directory on the delay buffer 1515 and The delayed video/audio is acquired from the delay buffer 1515 for the network address of the application/game server 1521-1525 from which the video/audio source is delayed. Once all such requests have been generated and processed, up to 15 thumbnail-sized live video streams will be routed from the outgoing route 1540 to the incoming route 1502, to the application/game that produces the user interface screen. Servers 1521-1525 will be decompressed and displayed by the server. The screen size of the delayed video/audio stream may be too large, and in this case, the application/game server 1521-1525 will decompress the streams and scale the video streams down to a thumbnail size. In one embodiment, the request for audio/video is sent to (and managed by) a central "management" service (not shown in FIG. 15) similar to the colocation service control system of FIG. 4a, the central " The Manage service then redirects the requests to the appropriate application/game server 1521-1525. Moreover, in an embodiment, no requests may be required as the thumbnails are "pushed" to the users of their users who are permitted to do so.

Mixing all of the audio from 15 games at the same time creates a harsh sound. The user can choose to mix all the sounds together in this way (perhaps only one of the "noisy" feelings produced by all the actions being viewed), or the user can choose to listen to only one game at a time. The selection of a single game is achieved by moving the yellow selection block 1601 (which appears as a black rectangular outline in the black and white diagram of Figure 16) to a given game (by using the arrow keys on the keyboard, by Move the mouse to move the yellow selection block by moving the joystick or by pushing a direction button on another device such as a mobile phone. Once a single game is selected, the audio from the game is played. Also show game information 1602. In the case of this game, for example, an issuer logo (for example, "EA" of "Electronic Arts") and a game logo (for example, "Need for Speed Carbon" and an orange horizontal column (reproduced as one in FIG. 16) The vertical strip)) indicates the number of people who play or watch the game at a particular time in the relevant item (in this case, there are many people, so the game is "Hot"). In addition, "Stats" (ie, statistics) is provided, which indicates that 145 players are actively playing 80 different execution individuals of "The Need for Speed Game" (ie, the game may be played by individual players or multiple players). Come play) and there are 680 viewers (this user is one of them). Please note that such statistics (and other statistics) are collected by the hosted service control system 401 and stored on the RAID arrays 1511-1512 for maintaining a log of the operations of the hosted service 210 and for appropriate Billing for users and paying for publishers who provide content. Some of the statistics are recorded by the service control system 401 due to actions, and some statistics are reported by the individual application/game servers 1521-1525 to the service control system 401. For example, the application/game server 1521-1525 executing this Game Finder application sends a message of when to watch the game (and when to stop watching the game) to the colocation service control system 401, so the server can be updated. Statistics on the number of games viewed. Some of these statistics are available for user interface applications such as this Game Finder application.

If the user clicks on one of the start buttons on their input device, they will see the thumbnail video in the yellow box zoom in to full screen size while continuing to play live video. This effect will be shown in the process in Figure 17. Please note that the size of the video window 1700 has increased. To implement this effect, the application/game server 1521-1525 requests the application/game server 1521-1525 executing the game to select a copy of one of the full-screen video streams of the game (displayed by the user) The resolution of device 422 is delivered to the server. The application/game server 1521-1525 executing the game notifies the shared hardware compressor 1530 that a copy of the thumbnail size of the game is no longer needed (unless another application/game server 1521-1525 requires this thumbnail) And then the server instructs the compressor to send a copy of one of the full screen sizes of the video to the application/game server 1521-1525 that scales the video. The resolution of the display device 422 of the user playing the game may be the same or different than the resolution of the display device of the user who zoomed in on the game. Moreover, the resolution of the display device 422 of other viewers of the game is the same or different (and may have different audio playback components, such as stereo or surround sound) than the resolution of the display device of the user who zoomed in on the game. Therefore, the shared hardware compressor 1530 determines whether a suitable compressed video/audio stream that satisfies the needs of the user requesting the video/audio stream has been generated, and if there is a suitable compressed video/audio stream, The shared hardware compressor notification routing 1540 then delivers a copy of the stream to the application/game server 1521-1525 that scales the video, and if there is no suitable compressed video/audio stream The shared hardware compressor compresses another copy suitable for the user's video and instructs the outgoing route to send the stream back to the incoming route 1502 and the application/game server 1521-1525 that scales the video. . The server that is now receiving one of the full screen versions of the selected video will decompress the video and gradually enlarge the video to full size.

Figure 18 illustrates how the screen looks after the game has been fully zoomed in to the full screen and the game is displayed at full resolution of the user's display device 422, as indicated by the image pointed to by arrow 1800. The application/game server 1521-1525 executing the game finder application will send a message that no longer needs thumbnails to other applications/game servers 1521-1525 that are providing thumbnails and will no longer watch other games. Send to the colocation service control server 401. At this time, the only display generated by the server is an overlay image 1801 at the top of the screen, which provides information and menu control to the user. Please note that as this game progresses, the audience has grown to 2,503. Because of the large number of viewers, the scope is limited to many viewers having display devices 422 having the same or nearly the same resolution (each application/game server 1521-1525 has the ability to scale video to fit the device).

Since the game shown is a multiplayer game, the user can decide to join the game at some point. The colocation service 210 may or may not allow users to join the game for a variety of reasons. For example, a user may have to pay for playing a game and therefore choose not to play, the user may not have enough level to join the particular game (eg, it is not competitive compared to other players), or the user's Internet The road connection may not have a low enough delay to allow the user to play the game (eg, there is no delay constraint on viewing the game, so a game played at a distance (actually, on another continent) can be viewed without using Worry about delays, but for a game to be played, the delay must be low enough for the user to (a) enjoy the game and (b) to synchronize with other players who may have a shorter delay connection. The application/game server 1521-1525, which has provided the Game Finder user interface for the user, will request the colocation service control server 401 to start (ie, locate and start) the appropriate group for playing a specific game. The application/game server 1521-1525 loads the game from the RAID arrays 1511-1512, and then the hosted service control server 401 will indicate the incoming route 1502 to control the control The signal is transmitted from the user to the application/game server hosting the game, and it will instruct the shared hardware compression 1530 to self-compress the video/audio switch from the application/game server hosting the Game Finder application. To compress video/audio from the application/game server that currently hosts the game. The Game Finder app/game service is out of sync with the vertical sync of the new app/game server hosting the game, and as a result, There may be a time difference between the two synchronizations. Since the shared video compression hardware 1530 will start compressing the video after the application/game server 1521-1525 completes a video frame, the first frame from the new server can be Completed before the full frame time of one of the old servers, which can be completed before the previous compressed frame is completed (for example, consider the transmission time 992 of Figure 9b: if uncompressed frame 3 963 is advanced half of the picture When the frame time is completed, this will impact the transmission time 992). In this case, the shared video compression hardware 1530 will ignore the first frame from the new server (eg, if 974 is ignored) Block 4 964), and the client 415 maintains an additional frame time from the last frame of the old server, and the shared video compression hardware 1530 will begin to compress the new application/game server from the game hosting the game. The next frame time video of the device. Visually, the conversion from one application/game server to another server will be seamless for the user. The colocation service control server 401 then notifies that The application/game server 1521-1525 hosting the Game Finder switches to an idle state until the server is needed again.

The user can then play the game. And, the special thing is that it feels like the game can be played immediately (because the game has been loaded from the RAID arrays 1511-1512 to the application/game server 1521-1525 at a rate of one billion bits per second), and The game is loaded with a work system onto a server that is fully suitable for the game. The operating system is fully configured for the game with the ideal driver, registered configuration (in the case of Windows), and no other An application that may compete with the operation of the game is executed on the server.

In addition, as the user progresses through the game, each of the segments of the game will be at a rate of one billion bits per second (ie, loaded with one billion bytes in 8 seconds) from the RAID array 1511- The 1512 is loaded into the server, and because of the large storage capacity of the RAID arrays 1511-512 (because the RAID array is a shared resource among many users, it can be very large and still cost effective), so it can be pre-computed Geometry settings or other game clip settings are stored on the RAID arrays 1511-1512 and loaded very quickly. In addition, because the hardware configuration and computing power of each application/game server 1521-1525 is known, pixel shaders and vertex shaders can be pre-computed.

Therefore, the game is launched almost immediately, the game will be executed in an ideal environment, and then the clips are loaded almost immediately.

However, in addition to these advantages, the user will be able to view (via the previously described Game Finder and other means) other people playing the game and decide whether the game is interesting and learn from others while the game is interesting. skill. Moreover, the user will be able to demonstrate the game immediately without having to wait for a long download and/or installation time, and the user will be able to play the game immediately (possibly playing the game for a small fee, or playing the game for a long time). Moreover, the user will be able to be on a Windows PC, a Macintosh, on a television, at home, while traveling, and even on a mobile phone with a sufficiently low latency wireless connection (although the delay will not be sufficient for viewing) Question) Play the game. Moreover, this process can be done completely without having physically owned a copy of the game.

As mentioned previously, the user may decide not to allow others to view their gameplay, allow their games to be viewed after a delay, allow their games to be viewed by selected users, or allow their users to watch their games. In any event, in one embodiment, the video/audio will be stored in the delay buffer 1515 for 15 minutes, and the user will be able to "reverse" and watch their previous gameplay and pause, play slowly, forward fast, etc. Just as a user can watch a TV with a digital video recorder (DVR). Although in this example the user is playing a game, the same "DVR" capability is available to the user when using an application. This ability can be helpful in reviewing previous work and other applications as described below. In addition, if the game is designed to have the ability to reverse the game state information so that the camera view can be changed, etc., this "3D DVR" capability is also supported, but this capability will require the game to be designed to support this capability. Use the "DVR" capability of the delay buffer 1515 to work with any game or application (of course, limited to video generated when using a game or application), but in the case of a 3D DVR capable game, the user can control A three-dimensional "fly through" of the previously played clip, and the delay buffer 1515 records the resulting video, and records the game state of the game segment. Therefore, a particular "roaming" is recorded as compressed video, but since the game state is also recorded, a different roaming may occur at a later time in the same segment of the game.

As described below, users on the colocation service 210 will each have a "User Page" on which the user can post information and other information about themselves. Among other things, users will be able to post their saved video clips from the gameplay. For example, if the user has overcome a particularly difficult challenge in one of the games, the user can "reverse" to just before the point at which the user has achieved great success in the game, and then instruct the colocation service 210 A video clip having a certain duration (for example, 30 seconds) is stored on the user's "user page" for viewing by other users. To implement this process, the user only needs to use the application/game server 1521-1525 to play the video stored in the delay buffer 1515 to the RAID array 1511-1512, and then on the user's "user page". The video clip is indexed.

If the game has 3DDVR capabilities (as described above), the game state information required by the 3DDVR can also be recorded by the user and can be used for the user's "user page".

The Game Finder application will be used when a game is designed to have a "bystander" other than the active player (ie, a user who can travel in the 3D world and observe the action without participating in the action). Can join the game as a bystander and player. From an implementation point of view, if the user is a bystander rather than an active player, there is no difference to the hosting system 210. The game is loaded onto the application/game server 1521-1525 and the user will control the game (eg, control a virtual camera watching the world). The only difference will be the user's gaming experience.

Multiple user cooperation

Another feature of the colocation service 210 is the ability of multiple users to collaborate while watching live video, even if the devices with great differences are used for viewing. This feature is useful when playing games and using applications.

Many PCs and mobile phones are equipped with video cameras and have the ability to perform instant video compression, especially when the image is small. In addition, it is not difficult to implement instant compression by using a small camera attachable to a television and one of many hardware compression devices that use software or use compressed video. In addition, many PCs and all mobile phones have a microphone and there is a headset with a microphone.

These cameras and/or microphones (in particular using the low-latency video compression techniques described herein) in combination with native video/audio compression capabilities will enable users to self-use video and/or audio with input device control data. The location 211 is transmitted to the colocation service 210. When using such techniques, one of the capabilities illustrated in Figure 19 can then be achieved: a user can have their video and audio 1900 appear on a screen in another user's game or application. This example is a multiplayer game where teammates collaborate in a car race. The user's video/audio is only selectively viewable/audible by his teammates. Moreover, since there may be no delays effectively using the techniques described above, the players will be able to talk or move in real time with each other without audible delay.

This video/audio integration is achieved by having compressed video and/or audio from a user's camera/microphone as incoming Internet traffic 1501. Next, the incoming route 1502 routes the video and/or audio to the application/game server 1521-1525, which permits viewing/listening to the video and/or audio. Then, the users of the respective application/game servers 1521-1525 (which choose to use the video and/or audio) decompress the video and/or audio and integrate them as needed to appear in the game or application. As explained in 1900.

The example in Figure 19 shows how to use this collaboration in a game, but this collaboration can be a very powerful tool for one of the applications. Consider the case of Chicago Architects designing a large building for New York City for one of New York’s property developers, but the decision involves a financial investor who is traveling and happens to be at Miami’s airport and must make a How to adapt to the decision-making of specific design elements in nearby buildings to meet both investors and property developers. Suppose that Chicago's construction company has a high-resolution monitor with a camera attached to the PC. The real estate developer in New York has a laptop with a camera, and the investor in Miami has a camera. mobile phone. The construction company can use the colocation service 210 to host a powerful architectural design application capable of highly realistic 3D rendering, and it can utilize a large database of buildings in New York City and a library of buildings in the design. The architectural design application will execute on one of the application/game servers 1521-1525 or on some of the servers (if the application requires a significant amount of computing power) to execute. Each of the three users in completely different locations will be connected to the colocation service 210, and each user will have a simultaneous view of the video output of the architectural design application, but by sharing the hardware Compression 1530 gives the simultaneous view the appropriate size for each user with a given device and network connectivity feature (eg, a building company can see a 2560x1440 60 fps display via a 20 Mbps commercial internet connection, New York Property developers can see a 1280x720 60 fps image on their laptop via a 6 Mbps DSL connection, and investors can see a 320x180 60 fps on their mobile phone via a 250 Kbps cellular connection. Image). Each party will hear the other party's voice (the conference call will be handled by any of the widely available conference call suite software in the application/game server 1521-1525) and via actuation of a user input device With one button, the user will be able to use their local camera to present their own video. As the meeting progresses, the architect will be able to demonstrate in a 3D representation of the photorealistic reality how the building looks when the architect rotates the building and is flying over another building in the area, and all The same video will be seen at the resolution of each side of the display device. There is no one of the native devices used by either party that can't handle the 3D animation with this realism, let alone download or even store the huge database needed to reproduce the surrounding buildings in New York City. From the standpoint of each of the users, despite the distance and even though the native devices are completely different, such users will have a seamless experience with an incredible degree of realism. And, when one wants their expression to be seen to better convey their emotional state, they can do the same. In addition, if a property developer or investor wants to control the building program and uses its own input device (which can be a keyboard, mouse, keypad or touch screen), then the input device can be used and the building program will There is no response delay in response (assuming that the network connection of the property developer or investor does not have an unreasonable delay). For example, in the case of a mobile phone, if the mobile phone is connected to the WiFi network of the airport, it will have a very low latency. However, if a mobile phone uses a cellular data network available in the United States today, it is likely to suffer significant lag. In addition, it should be acceptable for investors to watch most of the conference goals of the architect's control over the building or for the video conference call, even if there is a honeycomb delay.

Finally, at the end of the collaborative meeting call, the property developer and investor will comment and stop the colocation service, and the construction company will be able to "reverse" the video of the meeting that has been recorded on the delay buffer 1515 and review the Comments, facial expressions and/or actions applied to the 3D model of the building. If there are specific segments that the construction company wants to store, then the segments of the video/audio can be moved from the delay buffer 1515 to the RAID arrays 1511-1512 for archival storage and later playback.

In addition, from a cost perspective, if an architect only has to use computing power and a large database in New York City for a 15-minute conference call, then it only needs to pay for the time to use the resources, without having to have high power. The workstation does not have to purchase an expensive copy of a large database.

Video service rich in video

The colocation service 210 implements an unprecedented opportunity to establish a video-rich community service on the Internet. 20 shows an exemplary "user page" of one of the game players on the colocation service 210. Like the Game Finder application, the "user page" is an application that executes on one of the application/game servers 1521-1525. All thumbnails and video windows on this page show continuous moving video (if the clip is shorter, it loops).

Using a video camera or by uploading video, the user (the user name is "KILLHAZARD") can also publish one of his own video 2000s that other users can view. The video is stored on the RAID arrays 1511-1512. In addition, when other users come to the KILLHAZARD "user page", if KILLHAZARD is using the colocation service 210 at this time, it will show the live video 2001 of what KILLHAZARD is doing (assuming it allows users to view it). User page" to view it. This process is implemented by the application/game server 1521-1525 hosting the "user page" application, which requests the service control system 401 whether the KILLHAZARD is active and if In effect, the application/game server 1521-1525 that KILLHAZARD is using is requested. Then, using the same method used by the Game Finder application, a compressed video stream with the appropriate resolution and format is sent to the application/game server executing the "user page" application and the script will be displayed. Compressed video stream. If a user selects the window of KILLHAZARD's live gameplay and then clicks appropriately on their input device, the window will zoom in (again using the same method as the Game Finder application) and the live video will be viewed The resolution of the user's display device 422 (suitable for viewing the characteristics of the user's Internet connection) is full of screens.

A key advantage of this prior art approach is that the user viewing the "user page" can see the live game that the user does not own, and most likely does not have a native computer or game console capable of playing the game. This advantage provides the user with a good opportunity to see the user "playing" the game shown in the "user page", and this is an opportunity to learn to watch a game that the user may want to try or improve.

A camera-recorded or uploaded video clip 2002 from a partner of KILLHAZARD is also displayed on the "user page" and under each video clip is a text indicating whether the partner is playing the game online (eg, six_shot is playing The game "Eragon" (shown here as Game4) and MrSnuggles99 offline, etc.). By clicking on a menu item (not shown), the partner video clips are switched from the display recorded or uploaded video to the partner currently playing the game on the colocation service 210. What is the live video of what? Therefore, it becomes the partner's Game Finder group. If a game of a partner is selected and the user clicks on it, the game will be zoomed in to the full screen, and the user will be able to watch the game played live on the full screen.

In addition, the user of the game watching the partner does not own the copy of the game or the local computing/game console resource to play the game. The game viewing is actually instantaneous.

As previously described above, when a user plays a game on the colocation service 210, the user can "reverse" the game and find a video segment that he wants to store and then store the video clip to it. User page. This and other actions "to show off clips ^TM." These video clips 2003 are all sports clips 2003 stored by KILLHAZARD from previous games that they have played. The number 2004 shows a number of times the show clip has been viewed, and when viewing the show clip, the user has a chance to evaluate it, and the orange (here shown as a black outline) keyhole shape icon number 2005 indicates the level. The level. These flaunting clips 2003 continue to circulate with the rest of the video on the page as the user views the "user page". If the user selects one of the show clips 2003 and clicks on it, the show clip zooms in to present the show clip 2003 and allows the clip to be played, paused, reversed, forward fasted, stepped Wait for DVR control.

The avatar clip 2003 play is implemented by the application/game server 1521-1525 loading the compressed video clips stored on the RAID arrays 1511-1512 when the user records the highlight clips and decompressing the show clips and playing them.

The show clip 2003 can also be a "3D DVR" video clip from a game that supports this capability (i.e., a sequence of game states from a game that can be replayed and allows the user to change the camera's point of view). In this case, game state information is also stored in addition to the compressed video recording of the particular "roaming" that the user made while recording the game segment. When viewing the "user page" and all thumbnails and video live loops continue, a 3D DVR show clip 2003 will continue to cycle through the smashing clips 2003 recorded as compressed video when the user logs "roam" of the game clip. However, when a user selects a 3D DVR to show off the clip 2003 and clicks on it, the user will be able to give a 3D of the game segment in addition to the DVR control that allows the compressed video to show off the clip. Click on the button of the DVR capability. The user will be able to independently control a camera "roaming" during the game segment, and if the user wishes (and has the user's permission for the user page), it will be able to record an alternative sports clip in the form of a compressed video. Roaming, which can then be used by other viewers of the user page (immediately, or after the owner of the user page has reviewed the opportunity to show off the clip).

This 3D DVR Show Clip 2003 capability is achieved by launching a game that will replay the recorded game state information on another application/game server 1521-1525. Since the game can be started almost instantaneously (as previously described), there is no difficulty in starting the game, where its playback is limited to the state of the game recorded by the show clips, and thus allows the user to record the compressed video to the delay buffer 1515. Use a camera to "roam". Once the user has completed "roaming", the game is revoked.

From the user's point of view, using a 3D DVR to show off the clip 2003 to launch a "roaming" is no more than controlling the DVR control of a linear show clip 2003. The DVR control may not know the game or even how to play the game. The DVR control is only a virtual camera operator staring at the 3D world during the recording of the game segment by the other.

The user will also be able to place their own audio on top of the show clip, which is recorded or uploaded from the microphone. In this way, a showcase clip can be used to generate custom animations using characters and actions from the game. This animation technique is often referred to as "machinima".

As the user progresses through the game, the users will reach different skill levels. The game played will report the completion to the service control system 401 and these skill levels will be displayed on the "user page."

Interactive animation

Online advertising has been converted from text to still images, video and now to interactive clips, which are typically implemented using an animated thin client like Adobe Flash. The reason for using an animated thin client is that the user is usually impatient with being delayed in order to fix the product or service with the special permissions that apply. In addition, the thin client is executed on very low performance PCs, and as a result, advertisers can have a high level of confidence that interactive ads will work properly. Unfortunately, animated thin-clients such as Adobe Flash are limited in terms of interaction and duration of experience (to reduce download time and be available on almost all consumer devices (including low-performance PCs and Macs without GPUs or high-performance CPUs) ) on the operation).

Figure 21 illustrates an interactive advertisement in which the user selects the exterior and interior colors of the car as it is rotated in a showroom, while instant ray tracing shows how the car looks. The user then selects an avatar to drive the car, and then the user can drive the car on the runway or in a country other than Monaco. The user can select a larger engine or a better tire and can then view the impact of the changed configuration on the acceleration and grip capabilities of the vehicle.

Of course, the ad is actually a perfect 3D video game. However, for this ad that can be played on a PC or a game console, it may require a 100 MB download, and in the case of a PC, it may require a special driver to be installed and the PC lacks sufficient CPU or GPU computing. In the case of ability, it cannot be executed at all. Therefore, such advertisements cannot be implemented in previous technical configurations.

In the colocation service 210, these advertisements begin almost immediately and execute flawlessly, regardless of the capabilities of the user client 415. As a result, these ads start faster, leaner, and more reliable than streamlined, user-interactive ads.

Streaming geometry during instant animation

The RAID arrays 1511-1512 and the inbound routing 1502 can provide data rates that are fast and have very low latency to potentially design dependent on the RAID arrays 1511-1512 and the incoming routing 1502 for instant animation (eg, one has During the roaming of a complex database, video games and applications that play the geometric shapes of the game or in an application are reliably transmitted.

With regard to prior art systems (such as the video game system shown in Figure 1), the available mass storage devices (especially in certain home devices) are too slow to stream geometry during game play (required geometry) Except for some degree of predictability). For example, in a driving game in which there is a prescribed road, the geometry of the building entering the field of view can be reasonably predicted, and the mass storage device can pre-search for the location of the upcoming geometry.

However, in a complex scene with unpredictable changes (for example, in a battlefield scene with complex characters around), if the RAM on a PC or video game system is completely filled with the geometry of the object that is currently seen, see The shape, and then the user suddenly turns his character to see behind his character, if the geometry has been preloaded into the RAM, there may be a delay before the geometry can be displayed.

In the colocation service 210, the RAID arrays 1511-1512 can stream data over one billion bit Ethernet speed, and for a SAN network, possibly via a 10 billion bit Ethernet or via other Network technology has reached a rate of 10 billion bits per second. 10 billion bits per second will load 1 billion bytes of data in less than a second. In a 60 fps frame time (16.67 milliseconds), approximately 170 million bits (21 MB) of data can be loaded. Rotating media (of course, even in a RAID configuration) will still cause a delay greater than one frame time, but the flash memory based on the flash memory will eventually be as large as the rotating media RAID array and will not cause this high latency. . In one embodiment, using a large amount of RAM to fully write to the cache provides very low latency access.

Therefore, with a high enough network speed and a sufficiently low latency mass storage, the geometry can be streamed to the application/game server 1521-1525 as fast as the CPU and/or GPU can process 3D data. . Thus, in the previously given example where the user suddenly turned his character and looked backwards, the geometry of all subsequent characters can be loaded before the character completes the rotation, and thus, for the user, Going up seems to be in a real-life world of photos that are as real as real scenes.

As discussed earlier, one of the final boundaries in a photorealistic computer animation is a human face, and due to the sensitivity of the human eye to the sputum, the tiniest error from a photo-realistic face can lead to negative reactions from the viewer. . Figure 22 shows how to use the Contour ^TM Reality Capture technology to capture one live performance (the same technology as the subject of the application in the application of: number ( "Apparatus and method for application of September 15, 2004 10 / 942,609 "Capturing the motion of a performer"); No. 10/942, 413, "Apparatus and method for capturing the expression of a performer", filed on September 25, 2004; ("Apparatus and method for improving marker identification within a motion capture system"); No. 11/077,628 ("Apparatus and method for performing motion capture using shutter synchronization"), March 10, 2005; October 20, 2005 Application No. 11/255, No. 11/449,131 ("System and method for performing motion capture using" on June 7, 2006 ("Apparatus and method for performing motion capture using a random pattern on capture surfaces") Phosphor application techniques"); No. 11/449,043, filed on June 7, 2006 ("System and method for performing motio n capture by strobing a fluorescent lamp");"System and method for three dimensional capture of Stop-motion animated characters", June 7, 2006, each of these applications Let the assignee of the CIP application generate a very smooth capture surface, and then generate a tracking surface with a high polygon count (i.e., the polygon motion accurately follows the motion of the face). Finally, when the video of the live performance is mapped onto the tracking surface to produce a textured surface, a photorealistic result is produced.

While current GPU technology is capable of instantly reproducing many of the polygons in the tracking surface and texture and illuminating the surface, if the polygons and textures change over time in each frame (which will produce the most photorealistic results), then the technique It will quickly exhaust all available RAM from a modern PC or game console.

Using the stream geometry techniques described above, the geometry is continuously fed into the application/game server 1521-1525 such that the servers can continuously animate the photo realistic face, thereby It has become practical to allow a video game to be formed with a face that is almost indistinguishable from a live face.

Integration of linear content and interactive features

Animation, television programming and audio materials (collectively referred to as "linear content") can be widely used by home and office users in many forms. Linear content is available on physical media such as CD, DVD and Blu-ray media. Linear content can also be recorded by the DVR from satellite and cable television broadcasts. Moreover, linear content may be obtained via satellite and cable television as pay-per-view (PPV) content or may be available as video-on-demand (VOD) on cable television.

More and more linear content is available via the Internet as both downloadable content and streaming content. Today, there really is no place to experience all the features associated with linear media. For example, DVDs and other video optical media often have interactive features that are not available elsewhere, such as director commentary, "production" specials, and the like. Online music sites have cover images and song information that are not normally available on CDs, but not all CDs are available online. Also, Web sites associated with television shows often have additional features, blogs, and sometimes comments from actors or creators.

In addition, for many animations or sports competitions, there are often video games that are often released with linear media (in the case of animation) or (in the case of sports games) that can be closely linked to real-world events (eg, player transactions). .

The colocation service 210 is well suited for delivering linear content with completely different forms of related content. There is no doubt that delivering animations is no more challenging than delivering highly interactive video games, and the colocation service 210 is capable of delivering linear content to many devices (in the home or office) or to mobile devices. 23 shows an exemplary user interface page of the colocation service 210 that shows the selection of linear content.

However, unlike most linear content delivery systems, the colocation service 210 can also communicate relevant interactive components (eg, menus and features on DVDs, interactive overlays on HD-DVDs, and Adobe Flash animations on Web sites). (as explained below)). Therefore, the client device 415 limits no further restrictions on which features are available.

In addition, the escrow system 210 is capable of dynamically and instantly linking linear content to video game content. For example, if a user watches a Quidditch game in the movie Harry Potter and decides that he or she is willing to try to play Quidditch, the user can click on only one button and the movie will pause and immediately turn the user Connect to the Quidditch episode of the Harry Potter video game. After participating in the Quidditch competition, press another button and the movie will start over immediately.

For photorealistic graphics and production techniques (where video captured by photography is difficult to distinguish from real-life characters), as described in this article, when a user converts from a Quidditch game in a real-life movie to a co-hosting When one of the video games is one of the Quidditch games, the two scenes are virtually indistinguishable. This is a new creative option for directors of linear content and interactive (eg video game) content, as the boundaries between the two worlds become indistinguishable.

Using the colocation service architecture shown in Figure 14, the viewer can be provided with control over the virtual camera in the 3D movie. For example, in a scene that occurs within a train compartment, the viewer may be allowed to control the virtual camera and look around the car as the story progresses. This assumes that all 3D objects ("assets") in the car are available and have sufficient computing power to instantly reproduce the scenes as well as the original movie.

And, even for non-computer-generated entertainment, there are exciting features that can be provided. For example, the 2005 animation "Pride and Prejudice" has many scenes in a gorgeous old-fashioned English manor. For a particular estate scene, the user can pause the video and then control the camera to explore the estate, or possibly the surrounding area. To carry out this process, a camera with a fisheye lens can be carried through the manor to remember its location, much like the Apple, Inc. QuickTime VR implementing the prior art. The various frames can then be transformed to undistort the images, and then the frames are stored with the movie on the RAID arrays 1511-1512 and played when the user chooses to perform a virtual tour.

Regarding the sports competition, a live sports game such as a basketball game can be streamed through the colocation service 210 for the user to watch, just as the user watches the regular television. After the user watches a particular game, one of the game's video games (which ultimately makes the basketball team player look like the real player) can catch up with the player starting at the same position, and the user (possibly each user controls) A player can replay the game to see if it can do better than the player.

The colocation service 210 described herein is extremely well-suited for supporting this future world because it can utilize computing power and can't actually install large-capacity storage resources in the home or most office environments, and the calculation of colocation services Resources are always up to date and the latest computing hardware is available, while in a home environment, families will always have older PCs and video games. Moreover, in the colocation service 210, all such computational complexity is hidden from the user, so even if the user may be using a very complicated system, it is as simple as changing the television channel from the user's point of view. In addition, such users will be able to gain all the computing power and computing power from any client 415 experience.

multiplayer game

In the case of a game for a multiplayer game, the game will be able to communicate to the application/game server 1521-1525 via the portal 1502 and utilize one of the Internet bridges (not shown) ) communicates with a server or game machine that is not executing in the colocation service 210. When playing multiplayer games on a computer over a typical network, the application/game server 1521-1525 will have the benefit of extremely fast access to the Internet (as opposed to executing games on one of the servers in the home) But these servers will be limited by the ability to play other games on the slower connection, and may also be limited by the fact that the game servers on the Internet are designed to accommodate the least commonalities, It can be a home computer connected to a relatively slow consumer Internet connection.

However, many differences can be achieved when playing a multiplayer game entirely within a hosted service 210 server center. Each application/game server 1521-1525 hosting a game of one user will be interconnected with other application/game servers 1521-1525 and any server hosting the central control of the multiplayer game, It has extremely high speed, very low latency connectivity and a large and extremely fast storage array. For example, if a billion-element Ethernet is used for the incoming route 1502 network, the application/game server 1521-1525 will communicate with each other and communicate to the escrow at a rate of one billion bits per second. Any server controlled centrally by this multiplayer game may have a delay of only 1 millisecond or less. In addition, the RAID arrays 1511-1512 will be able to respond very quickly and then transfer data at a rate of one billion bits per second. As an example, if a user customizes a character based on appearance and equipment so that the character has a large number of geometries and behaviors that are unique to the character, the prior art system is limited to being executed at home on a PC or game console. The game client, if the character will enter the view of another user, the user will have to wait for a long and slow download to complete, causing all geometry and behavioral data to be loaded into their computer. In the colocation service 210, the same download can be servoed at a rate of one billion bits per second from the RAID arrays 1511-1512 on a one billion bit Ethernet. Even if the home user has an 8 Mbps internet connection (which is extremely fast for today's standards), the one billion bit Ethernet is 100 times more connected. Therefore, a one-minute action on a fast internet connection can take no more than one second on a billion-bit Ethernet.

Top player grouping and competition

The colocation service 210 is well suited for competition. Since no game is executed in one end user, the user has no chance to cheat (for example, the user may modify a game executed on his own PC to give him an improper benefit in a prior art competition) Cheating in the middle). Moreover, due to the ability of the output route 1540 to multicast UDP streams, the colocation service 210 can simultaneously broadcast important contests to thousands or more viewers.

In fact, when there is such a popularity that thousands of users receive a particular video stream of the same stream (eg, showing a view of an important race), the video stream can be sent more efficiently to a Akamai or Limelight's Content Delivery Network (CDN) is widely distributed to many client devices 415.

A similar level of efficiency is obtained when using a CDN to display the Game Finder page of the top player group.

For important competitions, a well-known announcer can be used to provide commentary during a specific competition. While many users will watch an important competition, relatively few users will participate in the competition. The audio from the famous announcer can be delivered to the application/game server 1521-1525 that hosts the user of the contest and hosts any of the viewer modes of the game in the contest, and can add the audio Recorded on the game audio. The video of a famous announcer can also be overlaid on the game and may only be overlaid on the bystander view.

Web page loading acceleration

The World Wide Web and its main transport protocol Hypertext Transfer Protocol (HTTP) are conceived and defined in the era of enterprises with high-speed Internet connectivity, and consumers on the line use dial-up modems or ISDN. At the time, the "gold standard" for fast connections was the T1 line, which symmetrically provided a data rate of 1.5 Mbps (i.e., equal data rates in both directions).

Today, the situation is completely different. In many developed countries, the average home connection speed via a DSL or cable modem connection is much higher than the data rate below the T1 line. In fact, in some parts of the world, fiber-to-the-curb brings data rates of up to 50 to 100 Mbps to the home.

Unfortunately, HTTP has not been constructed (and not implemented) to effectively utilize these significant speed improvements. The website is a collection of files on a remote server. In a very simple condition, HTTP requests the first file, waits for the file to be downloaded, and then requests the second file, waiting to download the file, and the like. In fact, HTTP allows more than one "open connection" (ie, more than one file) to be requested at a time, but only a few open connections are allowed due to agreed standards (and the need to prevent overloading of the web server). In addition, due to the way web pages are constructed, browsers often do not know multiple simultaneous pages that can be downloaded immediately (ie, only after parsing a page, it is necessary to download a new file (such as an image)). Therefore, the files on the website are essentially loaded one by one. Moreover, due to the request response protocol used by HTTP, there is a typical 100 millisecond delay associated with each file loaded (accessing a typical web server in the United States).

For relatively low speed connections, this does not introduce a number of problems, as the download time of the files themselves accounts for the majority of the waiting time for such web pages. However, as connection speeds increase, especially for complex web pages, problems begin to emerge.

In the example shown in Figure 24, a typical commercial website (this particular website is from a major athletic shoe brand) is shown. There are 54 files on this website. These files include HTML, CSS, JPEG, PHP, JavaScript, and Flash files, and include video content. A total of 1.5 MB must be loaded before the live broadcast of the page (ie, the user can click on the page and start using the page). There are many reasons for accessing a large number of files. First, the page is a complex and complete web page, and secondly, the page is based on information about the user accessing the page (eg, which country the user is from, which language is used, whether the user has previously purchased Etc.) Dynamically combine web pages and download different files depending on all of these factors. In addition, the page is a very typical commercial web page.

Figure 24 shows the amount of time that elapses before the live broadcast of the web page as the connection speed increases. For a connection speed of 2,500 Mbps, a conventional web server with a conventional web browser is used, and it takes 13.5 seconds until the web page is live. For a connection speed of 2402 at 12 Mbps, the load time is reduced to 6.5 seconds, or approximately doubled. However, for a connection speed of 2403 at 96 Mbps, the load time is only reduced to approximately 5.5 seconds. The reason is that at such a high download speed, the time to download the files themselves is the smallest, but the delay of each file (about 100 milliseconds per file) still exists, resulting in 54 files * 100 milliseconds = 5.4 seconds delay . Therefore, no matter how fast the connection to the home is, the website will always take at least 5.4 seconds before the live broadcast of the webpage. Another factor is the server side row; each HTTP request is added after the queue, so on a busy server, this will have a significant impact because of every small item that will be obtained from the web server. HTTP requests must wait for their timing.

One way to solve these problems is to drop or redefine HTTP. Or, it might be possible for site owners to better merge their files into a single file (for example, in Adobe Flash format). However, as a practical matter, this company and many other companies have invested heavily in their website architecture. In addition, although some homes have a 12-100 Mbps connection, most homes still have slower speeds, and HTTP does work well at slow speeds.

An alternative method is to host a web browser on the application/game server 1521-1525, and on the RAID arrays 1511-1512 (or, if possible, the application/game server 1521 hosting the web browsers) The files on the server of the website are hosted on the RAM in the -1525 or on the local storage. Due to the extremely fast interconnection via the incoming route 1502 (or to the local storage), there is not a 100 millisecond delay for each file in the case of using HTTP, and there will be a delay in which each file can be ignored in the case of using HTTP. . Then, instead of the user accessing the webpage via HTTP in his home, the user can access the webpage via the client 415. Then, even with a 1.5 Mbps connection (because the video of this web page does not require a large bandwidth), the web page will be live broadcast in less than 1 second (according to line 2400). In essence, there is no delay before the web browser executing on the application/game server 1521-1525 displays a live broadcast page, and there is no detectable delay before the user terminal 415 displays the video output from the web browser. . When the user slides the mouse around the webpage and/or type on the webpage, the user's input information will be sent to the web browser executing on the application/game server 1521-1525, and the web browser will respond accordingly. Respond to the ground.

One disadvantage of this method is that if the compressor continues to transmit video data, the bandwidth is used even if the web page becomes static. This disadvantage can be remedied by configuring the compressor to only transmit data when the web page changes (and if the web page changes) and then only transfer the data to portions of the changed page. While there are some web pages with constantly changing blinking banners, etc., such web pages tend to be annoying, and unless there is a reason to move something (eg, a video clip), the web page is typically static. For such web pages, the situation may be as follows: the information transmitted using the colocation service 210 is less than that transmitted using a conventional web server, since only the actual displayed image will be transmitted, and the reduced client executable code will not be transmitted. It does not transmit large objects that may never be viewed, such as scrolling images.

Therefore, using the colocation service 210 to host the old webpage, the webpage loading time can be shortened to the case where opening a webpage is like changing a television channel: the webpage is effectively and effectively broadcasted immediately.

Promote debugging of games and applications

As mentioned earlier, video games and applications with instant graphics are very complex applications, and in general, when such applications are released to the market, such applications contain program errors. While software developers are able to get feedback on bugs from users, and such software developers may have some means to restore machine state after a crash, accurately identify a game or instant application that is corrupted The reason for the wrong execution is very difficult.

When a game or application is executed in the colocation service 210, the video/audio output of the game or application is continuously recorded on the delay buffer 1515. In addition, a watchdog handler is executed on each application/game server 1521-1525, which periodically reports the application/game server 1521-1525 smoothly to the hosted service control system 401. carried out. If the watchdog handler fails to report, the server control system 401 will attempt to communicate with the application/game server 1521-1525, and if successful, the server control system 401 will collect any available machine state. Any available information is sent to the software developer along with the video/audio recorded by the delay buffer 1515.

Thus, when the game or application software developer gets a notification of corruption from the colocation service 210, it begins to record the cause of the corruption on a frame-by-frame basis. This information can be very useful when tracking bugs and fixing them.

Please also note that when the application/game server 1521-1525 is corrupted, the server is restarted at the most recent restartable point and a message of technical difficulty is provided to the user.

Resource sharing and cost savings

The systems shown in Figures 4a and 4b provide various benefits to end users and game and application developers. For example, in general, home and office client systems (eg, PCs or game consoles) are only in use for a small percentage of a week. According to the "Active Gamer Benchmark Study" published by Nielsen Entertainment on October 5, 2006 (http://www.prnewswire.com/cgl-bin/stories.pl?ACCT=104&STORY=/www/story/10-05 -2066/0004446115&EDATE=), the active player spends an average of 14 hours a week playing video game console and spends about 17 hours a week playing handheld devices. The report also states that for all gaming activities (including consoles, handhelds, and PC games), active players spend an average of 13 hours per week. Considering the higher number of times you play console video games, there are 24*7=168 hours per week. This implies that in a player's home, a video game console is only 17/168=10% per week. The time is in use. Or, in 90% of the time, the video game console is idle. Given the high cost of video game consoles and the fact that manufacturers subsidize such devices, this is an extremely inefficient use of expensive resources. PCs in the enterprise are usually only used in a small part of the week, especially non-portable desktop PCs often require high-end applications such as Autodesk Maya. While some businesses are open all the time and on holidays, and use PCs at all times and during the holidays (for example, portable computers that take home to work at night), most corporate activities tend to focus on 9AM to 5PM Monday to Friday (in a given corporate time zone), a small number of holidays and rest periods (such as lunch), and since most PC usage occurs when the user actively uses the PC, it is determined that the table Type PC utilization tends to be after these business hours. If you assume that you continue to use your PC from 9AM to 5PM on a Friday, this will imply that you will use your PC for 40/168=24% of the week. High-performance desktop PCs are a very expensive business investment, and this reflects a very low level of utilization. Schools that teach on a desktop computer may use the computer in less time of the week, and although it varies depending on the teaching time, most of the teaching takes place during the daytime hours of Monday through Friday. Therefore, in general, PCs and video game consoles are utilized only for a fraction of the week.

It is worth noting that because many people work in business or school during the daytime from Monday to Friday during non-holiday, these people usually do not play video games during these time periods, and therefore, these people are usually at other times ( Play video games such as at night, on weekends or on holidays.

Given the configuration of the colocation service shown in Figure 4a, the usage patterns described in the last two paragraphs lead to extremely efficient use of resources. Obviously, there is no limit to the number of users that can be served by the colocation service 210 at a given time, especially if the user needs an immediate response to a complex application such as a sophisticated 3D video game. However, unlike video games used in the home or PCs used by businesses (which are typically idle most of the time), the server 402 can be reused by different users at different times. For example, a high-performance server 402 with high-performance dual-CPU and dual GPUs and a large amount of RAM can be used by enterprises and schools in non-holiday 9AM to 5PM, but it is played by players who play a perfect video game at night and on weekends. Used during holidays. Similarly, low-performance games can be used by businesses and schools during low-performance servers 402 with Celeron CPUs, no GPUs (or a very low-end GPU), and limited RAM during business hours, and a low-performance game can be used. The low performance server 402 is used during business hours.

In addition, for the hosted service configuration described in this article, resources will be effectively shared among thousands (if not millions) of users. In general, online services only have a small percentage of the total user base that uses the service at a given time. If we consider the Nielsen video game usage statistics listed previously, it is easy to understand why. If the player plays only 17 hours of console games a week, and if we assume that the peak usage time of the game is at night (5-12AM, 7*5 days = 35 hours/week) and weekends (8AM-12AM, 16*) 2=32 hours/week) Typical non-working, non-business hours, for a 17-hour game play time, there is a peak time of 35+32=65 hours a week. For a number of reasons, accurate peak user load on the system is difficult to estimate: some users will play games during off-peak hours, there may be specific daytimes when the user clusters peak, and the peak period can be affected by the type of game being played (eg It is very likely to play a children's game earlier in the evening). However, if the average number of game hours for a player is much less than the number of hours a day a player may play a game, then only a small portion of the number of users of the hosted service 210 will use the service at a given time. For this analysis, we will assume a peak load of 12.5%. Therefore, only 12.5% of the computation, compression, and bandwidth resources are used at a given time, so that only 12.5% of the hardware cost supports a given user to play a game of a given performance level, which is attributed to the resource. reuse.

In addition, if some games and applications require more computing power than other games and applications, resources can be dynamically allocated based on the game played by the user or the application executed by the user. Therefore, a user who chooses a low-performance game or application will be assigned a low-performance (less expensive) server 402, and a user who selects a high-performance game or application will be assigned a high performance (more Expensive) server 402. In fact, a given game or application can have a low-performance portion and a high-performance portion of the game or application, and the user can switch from one server 402 to another between the portions of the game or application. The 402 is executed on the lowest cost server 402 that maintains the user's needs in meeting the game or application. Note that RAID array 405 (which is much faster than a single disk) can even be utilized by low performance server 402, which has the benefit of faster disk transfer rates. Thus, the average cost per server 402 on all played games or applications used is much less than the cost of the most expensive server 402 playing the highest performance game or application, even the low performance server 402 will RAID array 405 yields disk performance benefits.

In addition, one of the servers 402 in the colocation service 210 may be only a PC motherboard that does not have a disk or peripheral interface except for a network interface, and may be integrated down to one of the SAN 403 sooner or later. One of the fast network interfaces is a single chip. In addition, the RAID array 405 is likely to be shared among many more users than the existing disk, so the cost of the disk for each active user will be much less than that of a disk drive. All of this device is likely to reside in a rack in an environment controlled server room environment. If the server 402 fails, the server can be easily repaired or replaced at the colocation service 210. In contrast, a PC or game console in a home or office must be a robust, self-contained electrical device that must withstand reasonable wear and tear from heavy blows or drops, require a housing, and have at least one magnetic Dish machines, which must withstand unfavorable environmental conditions (eg, be inserted into a superheated AV case with other gears), require a service guarantee, must be packaged and shipped, and are sold by retailers that have access to retail profits. In addition, even if you can play a low-performance game or application most of the time, a PC or game console must be configured to meet the peak of the most computationally intensive game or application that will be used at some point in the future. efficacy. Moreover, repairing a PC or console is an expensive and time consuming process (adversely affecting manufacturers, users, and software developers) if the PC or console fails.

Thus, if the system shown in Figure 4a provides the user with an experience comparable to the experience of a computing resource, then a user who experiences a given level of computing power in a home, office or school That said, providing this computing power via the architecture shown in Figure 4a is much cheaper.

Eliminate upgrade requirements

In addition, users no longer have to worry about upgrading PCs and/or consoles to play new games or deal with higher performance new applications. Regardless of the type of server 402 required for the game or application, any game or application on the colocation service 210 can be used by the user, and all games and applications are executed almost immediately (eg, quickly from the RAID array 405) Or the local storage on the server 402 is loaded and appropriately updated with the latest updates and program errors (ie, the software developer will be able to select an ideal server group for the server 402 executing a given game or application). The server 402 is then configured with the best driver, and over time, the developers will be able to provide updates, program error corrections, etc. to all copies of the game or application in the colocation service 210. ). In fact, after the user starts using the colocation service 210, the user may find that the game or application continues to provide a better experience (eg, via update and/or program error correction), and the situation may be as follows: After a year, it was discovered that a new game or application could be used on a service 210 that utilized computing technology that did not even exist a year ago (eg, a higher performance GPU), so it was impossible for the user to purchase it a year ago. The game can be played or executed after one year. Since the user does not see the computing resources for playing the game or executing the application (that is, from the user's point of view, the user only selects a game or application that is almost immediately executed, much like the user has changed the TV channel). Therefore, the user's hardware may have been "upgraded" and the user is not aware of the upgrade.

Eliminate backup needs

Another major issue for businesses, schools and home users is backup. Information stored in the local PC or video game console (for example, in the case of a console, the user's game scores and levels) can be lost in the event of a disk failure or inadvertent erasure. There are many applications available for manual or automatic backup of PCs, and the game console status can be uploaded to an online server for backup, but the local backup is usually copied to another local disk (or other non-volatile) Semaphores) must be stored in other secure locations and organized, and backups of online services are often limited by the slow upload speeds that are available via typical low-cost Internet connections. For the colocation service 210 of FIG. 4a, the data stored in the RAID array 405 can be configured using prior art RAID configuration techniques well known to those skilled in the art such that if a disk fails, no data is lost. A technician at the server center hosting the failed disk will be notified and the technician will replace the disk, which will then be automatically updated so that the RAID array can tolerate another failure. In addition, since all the drives are close to each other and have a fast local area network via the SAN 403, it is not difficult to configure all the disk systems in a server center to periodically back up the auxiliary storage. The auxiliary storage can be Stored in the center of the server or relocated outside of it. From the point of view of the user of the colocation service 210, the information has been very secure and it never needs to consider backup.

Access demonstration

Users often want to try a game or application before buying a game or application. As previously described, there are prior art techniques for demonstrating games and applications (the "demonstration" verb form means a trial version, which is also referred to as "demonstration" but as a noun), but Each is subject to restrictions and/or inconvenience. Using the colocation service 210, the user can easily and conveniently try out the demonstration. In fact, what the user has to do is to select and demonstrate the demonstration via a user interface, such as the user interface described below. The demonstration will be loaded almost immediately onto the server 402 suitable for the demonstration, and the demonstration will be performed like any other game or application. Whether the demonstration requires a very high performance server 402 or a low performance server 402, and regardless of the type of home or office client 415 used by the user, the demonstration will operate from the user's point of view. . The software publisher of the game or application demonstration will be able to accurately control which model and trials the user is allowed to try. Of course, the demonstration may include providing the user with a full version of the game or application that can be accessed. User interface component of opportunity.

Since the demonstration may be offered at a lower cost or free of charge, some users may attempt to use the demonstration over and over again (especially repeating the gameplay that may be interesting). The colocation service 210 can use various techniques to limit the exemplary use of a given user. The most straightforward method is to create a user ID for each user and limit the number of times a given user ID is allowed to play a demonstration. However, a user can set multiple user IDs, especially if they are idle. One technique for addressing this problem is to limit the number of times a given client 415 is allowed to play a demonstration. If the client is a standalone device, the device should have a serial number, and the colocation service 210 can limit the number of times a client with the serial number can access an exemplary number. If the client 415 is implemented as a software on a PC or other device, a serial number can be assigned by the colocation service 210 and stored on the PC to limit the exemplary use, but if the PC can be reprogrammed by the user and the serial number Erasable or changeable, another option for the colocation service 210 is to record the PC network adapter media access control (MAC) address (and/or other machine specific identifiers, such as hard drive serial numbers, etc.) And the exemplary use is limited to this. However, this method is not a very safe method if the MAC address of the network adapter can be changed. Another method is to limit the number of times a given IP address can be played. Although IP addresses can be periodically reassigned by cable modems and DSL providers, this does not happen often in practice, and if it can be determined (for example, by contacting an ISP) IP is accessed in a residential DSL or cable modem. In one of the IP addresses, a small amount of exemplary use can usually be arranged for a given household. In addition, there may be multiple devices in a home based on a NAT router that shares the same IP address, but typically in a residential environment, there are a limited number of such devices. If the IP address is in a block of a servo enterprise, a large number of demonstrations can be arranged for a company. However, in the end, all of these previously mentioned methods are combined to be the best way to limit the number of demonstrations on the PC. While there may not be a very safe way to limit the number of times a certain and skilled user repeats the demonstration, a large number of obstacles can create a sufficient barrier to abuse the demonstration system for most PC users. It's worth it, and the truth is that users will use these demos when they want to use the demo: try out new games and apps.

Benefits for schools, businesses and other institutions

The significant benefits of businesses, schools and other institutions that utilize the system shown in Figure 4a have increased in particular. Businesses and schools have significant costs associated with installing, maintaining, and upgrading PCs, especially when PCs suddenly execute high-performance applications such as Maya. As explained previously, PCs are typically utilized only for a fraction of the week, and when in the home, PCs with a given level of performance capability spend more in an office or school environment than in a server-centric environment. Much higher.

In the case of larger enterprises or schools (for example, large-scale universities), it is feasible for the IT department of such entities to set up a server center and maintain a remote access via a LAN-level connection. There are many solutions for remote access to a computer via a LAN or via a private high frequency wide connection between offices. For example, using Microsoft's Windows Terminal Server, or via a virtual network computing application (such as VNC from RealVNC, Ltd.), or through a streamlined client-side component from Sun Microsystems, users can access PCs remotely. Or a server with multiple quality graphical response times and user experience. In addition, such self-management server centers are typically dedicated to a single enterprise or school, and as such, the server centers cannot utilize the overlap of usage, which is in completely different applications (eg, entertainment applications and enterprise applications). It is possible to utilize the same computing resources at different times of the week. As a result, many businesses and schools lack the scale, resources, or expertise to independently set up a server center with a LAN speed network connection to each. In fact, a large percentage of schools and businesses have the same Internet connection as their homes (eg, DSL, cable modems).

However, such organizations may still require extremely efficient calculations on a regular or periodic basis. For example, a small construction company may have only a small number of architects and relatively modest computing needs in design work, but the company may periodically require extremely efficient 3D computing (eg, when generating a client for one) When the new building is designed for 3D roaming). The system shown in Figure 4a is well suited for such organizations. These organizations only need to provide the same kind of network connection (eg, DSL, cable modem) to the home and usually very cheap. Such organizations may utilize inexpensive PCs as the client 415, or completely eliminate the PC and utilize inexpensive dedicated devices that only implement control signal logic 413 and low latency video decompression 412. These features are particularly attractive to schools that may have PC theft or damage to specialized components in the PC.

This configuration addresses many of the problems of these organizations (and many of these advantages can also be shared by home users who perform general purpose computing). For example, operating costs (which ultimately must be passed back to the user in some form to have a viable business) can be much lower because (a) computing resources and other applications with different peak usage times during the week Sharing, (b) such organizations may access high-performance computing resources only when needed (and the cost of using high-performance computing resources), (c) such organizations do not have to provide for backup or otherwise maintain high-performance computing Resources of resources.

Eliminate piracy

In addition, games, apps, interactive movies, etc. are no longer being pirated as they are now. Because each game is stored on and executed on the colocation service 210, the user does not have the ability to access the basic code, so nothing is pirated. Even if a user can copy the original code, the user cannot execute the code on a standard game console or home computer. This opens up markets in the world, such as China, where standard video games are not available. It is also impossible to sell old games because there is no copy of the game that is distributed to the user.

For game developers, there are fewer market breaks when bringing a new generation of game consoles or PCs to market, which is what is happening today. Compared to the current situation (where a new generation of console or PC technology forces users and developers to upgrade and game developers rely on the timely provision of hardware platforms to users) (for example, in the case of PlayStation 3, its introduction After a delay of more than one year, and the developer has to wait until it is available and purchases a large number of units, the colocation service 210 can be gradually updated with more advanced computing techniques as the game needs change.

Streaming interactive video

The above description provides a number of applications that can be implemented by the novel basic concepts of low latency streaming interactive video (which implicitly includes audio and video, as used herein) based on the general Internet. Prior art systems that have been provided with streaming video over the Internet can only implement applications that can be implemented with high latency interaction. For example, basic playback controls for linear video (eg, pause, reverse, forward fast forward) work properly at high latency and may be selected in a linear video feed. Moreover, as previously explained, the nature of some video games allows for the playing of such video games in high latency situations. However, the high latency (or low compression ratio) of prior art methods for streaming video has severely limited the potential application of streaming video or narrowed its deployment to a dedicated network environment, and even in such environments, previously Technology also introduces a substantial burden on the Internet. The techniques described herein enable many applications via the Internet that may have low latency streaming interactive video, particularly for applications implemented via consumer Internet connections.

In fact, since the client device as small as the client 465 of Figure 4c is sufficient to provide an enhanced user experience, it has an actual arbitrary amount of computing power, an arbitrary amount of fast memory and a pole between powerful servers. Fast network connectivity, so it creates a new era of computing. In addition, because bandwidth requirements do not increase as the computing power of the system grows (ie, because bandwidth requirements are only dependent on display resolution, quality, and frame rate), broadband Internet connectivity is ubiquitous (eg, The problem is that complex users (such as PCs or mobile phones running Windows, Linux, OSX, etc.) or even complex applications (such as Windows, Linux, OSX, etc.) or even when they are reliable and have sufficient high bandwidth to meet the needs of all users' display devices 422 Whether a thin client (such as Adobe Flash or Java) is required for typical consumer and enterprise applications.

The emergence of streaming interactive video has led to reflections on the assumptions about the structure of the computing architecture. An example of such a structure is the colocation service 210 server center embodiment shown in FIG. The video path of the delay buffer and/or group video 1550 is a feedback loop in which the multicast stream video output of the application/game server 1521-1525 is output via path 1552 or immediately after a selectable delay. Path 1551 is fed back into the application/game server 1521-1525. This implementation is through many practical applications (eg, applications such as those illustrated in Figures 16, 17, and 20) that are not or are not achievable via prior art servers or local computing architectures. However, as a more general architectural feature, feedback loop 1550 provides for the retransmission of the streaming interactive video hierarchy because the video can be returned indefinitely as the application needs it. This achieves many application possibilities that have never been reached before.

Another key architectural feature is that the video stream is a one-way UDP stream. This effectively achieves any degree of multicasting of streaming interactive video (as compared to the increase in the number of users, two-way streaming such as TCP/IP streaming will cause more and more on the network due to back and forth communication. Traffic congestion). Multicasting is an important capability in the server center because multicasting allows the system to respond to the growing demand for Internet users (and, indeed, the world's population) to make one-to-many or even many-to-many communications. In addition, the examples discussed herein (such as Figure 16 illustrating the use of streaming interactive video retransmissions and multicasting) are only one of the giant icebergs of possibility.

Non-transfer level

In one embodiment, the colocation service 210 has one or more peer connections to one or more Internet Service Providers (ISPs) that also provide Internet services to the user, and in this manner, the host The escrow service 210 may be able to communicate with the user via a non-transit route that is reserved in the ISP's network. For example, if the colocation service 210 WAN interface 441 is directly connected to the network of Comcast Cable Communications, Inc., and provides a broadband service for the user premises 211 by using a Comcast cable modem, the colocation service 210 and the user A route between the ends 415 can be completely established in the Comcast network. Potential benefits of this approach include lower communication costs (because IP transfer costs between two or more ISP networks can be avoided), and a potentially more reliable connection (a congestion or other transfer between ISP networks) In the case of an interruption) and lower latency (in the case of congestion, inefficient routing or other delays between ISP networks).

In this embodiment, when the client 415 initially contacts the colocation service 210 at the beginning of a session, the colocation service 210 receives the IP address of the user premises 211. The colocation service then uses an available IP address table (eg, from the American Registry for Internet Numbers (ARIN)) to see if the IP address is assigned to a connected to colocation service. The IP address of a particular ISP 210 (which can be routed to the user premises 211 without IP forwarding for access via another ISP). For example, if the IP address is between 76.21.0.0 and 76.21.127.255, the IP address is assigned to Comcast Cable Communications, Inc. In this example, if the colocation service 210 maintains a connection to Comcast, AT&T, and Cox ISP, the colocation service selects Comcast as the ISP that is most likely to provide an optimal route to a particular user.

Video compression using feedback

In an embodiment, the feedback is provided from the client device to the colocation service to indicate successful (or unsuccessful) image block and/or frame delivery. The video compression operation of the colocation service is then adjusted using feedback information provided from the client.

For example, Figures 25a-25b illustrate an embodiment of the present invention in which feedback channel 2501 is established between client device 205 and colocation service 210. The feedback channel 2501 is used by the client device 205 to transmit a packetized response to the successfully received image block/frame and/or an indication of the unsuccessfully received image block/frame.

In one embodiment, after successfully receiving each image block/frame, the client transmits a response message to the colocation service 210. In this embodiment, if the colocation service 210 does not receive a response after a specified period of time, and/or if the colocation service 210 receives a response: the client device 205 has received a message that has not been sent. The subsequent escrow/frame of the image block/frame, the host escrow service 210 detects a packet loss. Additionally or alternatively, the client device 205 can detect that the packet is lost and will transmit an indication of one of the packet loss and an indication of one of the image blocks/frames affected by the packet loss to the colocation service 210. In this embodiment, there is no need for a continuous response to the successfully transmitted image block/frame.

Regardless of how a packet loss is detected, in the embodiment illustrated in Figures 25a through 25b, after generating an initial set of one of the I image blocks of an image (not shown in Figure 25a), the encoder then only generates P Image block until a packet is detected to be lost. Note that in Figure 25a, each frame (such as 2510) is illustrated as 4 vertical image blocks. The frame image may be blockped in a different configuration (such as 2x2, 2x4, 4x4, etc.), or the frame may be encoded as a whole without image blocks (ie, encoded as a large image) Piece). The foregoing examples of frame image block configurations are provided for illustrating this embodiment of the invention. The basic principles of the invention are not limited to any particular frame image block configuration.

Due to all of the reasons stated above (i.e., the P-picture block is typically smaller than the I-picture block), only transmitting the P-picture block reduces the bandwidth requirement of the channel. When a packet loss is detected via feedback channel 2501, a new I image block (as illustrated in Figure 25b) is generated by encoder 2500 to reinitialize the state of decoder 2502 on client device 205. As illustrated, in one embodiment, the I image blocks are spread across a plurality of encoded frames to limit the bandwidth consumed by each individual encoded frame. For example, in Figure 25 (where each frame includes 4 image blocks), a single I image block is transmitted at a different location in one of the four consecutive encoded frames.

Encoder 2500 can combine the techniques described with respect to this embodiment with other encoding techniques described herein. For example, in addition to generating an I-picture block in response to a detected packet loss, the encoder 2500 may generate an I-picture block in other cases, where the I-picture block may be beneficial for correctly reproducing the sequence of images (such as responding In the sudden scene conversion).

Figure 26a illustrates another embodiment of the present invention that relies on a feedback channel 2601 between the client device 205 and the colocation service 210. The encoder 2600 of this embodiment adjusts the dependency of the P block/frame in response to a detected packet loss resulting in a new I block/frame. As a primary matter, it should be noted that the specific details set forth in this example are not required in accordance with the basic principles of the invention. For example, although this example will be described using P-blocks/frames, the underlying principles of the invention are not limited to any particular encoding format.

In FIG. 26a, the encoder 2600 encodes a plurality of uncompressed image blocks/frames 2605 into a plurality of P blocks/frames 2606 and blocks the P blocks via a communication channel (eg, the Internet). The frame is transmitted to a client device 205. The decoder 2602 on the client device 205 decodes the P blocks/frames 2606 to generate a plurality of decompressed image blocks/frames 2607. The past state 2611 of the encoder 2600 is stored in the memory device 2610 on the colocation service 210, and the past state 2621 of the decoder 2602 is stored in the memory device 2620 on the client device 205. The "state" of the decoder is a well-known term in video coding systems such as MPEG-2 and MPEG-4. In one embodiment, the past "states" stored in the memory include combined data from previous P-image blocks/frames. The memories 2611 and 2621 can be integrated in the encoder 2600 and the decoder 2602, respectively, instead of being separated from the encoder 2600 and the decoder 2602, as shown in Fig. 26a. In addition, various types of memory may be used, by way of example and not limitation, and such memory includes random access memory.

In one embodiment, when no packet loss occurs, the encoder 2600 encodes each P image block/frame in dependence on the previous P image block/frame. Thus, as used in FIG. 26a as indicated by the notation, the image block P / P frame 4 dependent on the image block / frame 3 (using the notation identifying _43); P image blocks / frame 5 dependent on the image P block / box 4 (identified using the notation _54); and an image block P / P frame 6 in the image dependent block / frame 5 (representing identification method _65). In this example, P image blocks / frame ₄₃ has been lost during transmission between the encoder 2602 and decoder 2600. This loss can also be communicated to the encoder 2600 in a variety of ways including, but not limited to, the above methods. For example, whenever the decoder 2602 successfully receives and/or decodes an image block/frame, this information can be communicated from the decoder 2602 to the encoder 2600. If the encoder 2600 does not receive an indication of a particular image block/frame that has been received and/or decoded after a period of time, the encoder 2600 will assume that the image block/frame was not successfully received. Additionally or alternatively, the decoder 2602 can notify the encoder 2600 when a particular image block/frame is not successfully received.

In one embodiment, regardless of how the missing image block/frame is detected, once the missing image block/frame is detected, the encoder 2600 uses the last image block that is known to have been successfully received by the decoder 2602/ The frame encodes the next image block/frame. In the example shown in Figure 26a, image blocks/frames 5 and 6 are not considered "successfully received" because image blocks/frames 5 and 6 cannot be decoded by the decoder due to the loss of image block/frame 4. 2602 is correctly decoded (ie, the decoding of image block/frame 5 depends on image block/frame 4 and the decoding of image block/frame 6 depends on image block/frame 5). Therefore, in the example shown in FIG. 26a, the encoder 2600 is dependent on the image block/frame 3 (the last successfully received image block/frame) to encode the image block/frame 7, rather than relying on the decoder 2602. Image block/frame 7 is encoded in image block/frame 6 that cannot be decoded correctly. Although not illustrated in Figure 26a, the image block/frame 8 will then be encoded dependent on the image block/frame 7, and the image block/frame 9 will be encoded dependent on the image block/frame 8, assuming no additional is detected. The packet is lost.

As described above, the encoder 2600 and the decoder 2602 maintain the past encoder state 2611 and the past decoder state 2621 in the memories 2610 and 2620, respectively. Thus, when encoding image block/frame 7, encoder 2600 retrieves the previous encoder state associated with image block/frame 3 from memory 2610. Similarly, memory 2620 associated with decoder 2602 stores at least the last known good decoder state (in this example, the state associated with P-image block/frame 3). Thus, decoder 2602 retrieves past state information associated with image block/frame 3 such that image block/frame 7 can be decoded.

As a result of the above techniques, an instant, low-latency interactive video can be encoded and the interactive video can be streamed using a relatively small bandwidth because the I-picture block/frame is never needed (the decoder is initialized at the beginning of the stream and Except for the encoder). In addition, although the video image generated by the decoder may temporarily include the defect caused by the lost image block/frame 4 and the image block/frames 5 and 6 (which cannot be correctly decoded due to the loss of the image block/frame 4) Distortion, but this distortion can be seen in a very short duration. In addition, if an image block (rather than a full video frame) is used, the distortion will be limited to a particular area of the rendered video image.

A method in accordance with an embodiment of the present invention is illustrated in Figure 26b. At 2650, an image block/frame is generated based on a previously generated image block/frame. At 2651, a missing image block/frame is detected. In one embodiment, the missing image block/frame is detected based on information communicated from the encoder to the decoder, as described above. At 2652, a next image block/frame is generated based on an image block/frame that is known to have been successfully received and/or decoded at the decoder. In one embodiment, the encoder generates a next image block/frame by loading the state associated with the successfully received and/or decoded image block/frame from the memory. Similarly, when the decoder receives a new image block/frame, the decoder decodes the image block/frame by loading the state associated with the successfully received and/or decoded image block/frame from the memory. .

In an embodiment, the next image block/frame is generated at the encoder based on the last image block/frame that was successfully received and/or decoded. In another embodiment, the resulting image block/frame is an I image block/frame. In yet another embodiment, the selection based on how many image blocks/frames are lost and/or the delay of the channel is based on a previously successfully received image block/frame to generate the next image block/frame or to the next The image block/frame is generated as an I frame. In the case where a relatively small number (eg, 1 or 2) of image blocks/frames are lost and the round-trip delay is relatively low (eg, 1 or 2 frame times), a P block is generated/ The frame may be optimal because the difference between the last successfully received image block/frame and the newly generated image block/frame may be relatively small. If several image blocks/frames are lost or the round-trip delay is high, it may be optimal to generate an I-picture block/frame because the last successfully received image block/frame is between the newly generated image block/frame and the newly generated image block/frame The difference can be huge. In one embodiment, an image block/frame loss threshold and/or a delay threshold is set to determine whether to transmit an I block/frame or transmit a P block/frame. If the number of lost image blocks/frames is below the image block/frame loss threshold and/or if the round trip delay is below the delay threshold, a new I image block/frame is generated; Otherwise, a new P block/frame is generated.

In an embodiment, the encoder always attempts to generate a P block/frame for the last successfully received image block/frame, and if during the encoding process, the encoder determines that the P block/frame is likely to be An I image block/frame is large (for example, if the encoder has compressed one-eighth of the image block/frame and the compressed size is greater than the average size of the previously compressed I image block/frame) One), the encoder will abandon the compression of the P image block/frame and will instead compress an I image block/frame.

If packet loss is rare, the above system using feedback reporting a discarded image block/frame typically results in a very slight interruption to the user's video stream because an image block/frame is interrupted by a lost packet. Substituted in a round trip time between the client device 205 and the colocation service 210, it is assumed that the encoder 2600 compresses the image block/frame in a short time. Moreover, since the compressed new image block/frame is based on one of the uncompressed video streams, the video stream does not lag behind the uncompressed video stream. However, if a packet containing the new image block/frame is also lost, this will result in a delay of at least two round trips (which is used to request and send another new image block/frame again), in many practical In this case, the delay will cause a significant interruption in one of the video streams. Therefore, it is important that the newly encoded image block/frame sent after the discarded image block/frame is successfully transmitted from the colocation service 210 to the client device 205.

In an embodiment, the forward error correction (FEC) coding technique (such as the coding techniques previously described and illustrated in Figures 11a, 11b, 11c, and 11d) is used to reduce the loss of newly coded image blocks/frames. The chance. If FEC encoding has been used when transmitting image blocks/frames, a stronger FEC code is used for the newly encoded image block/frame.

One of the discarded packets may be due to a sudden loss of channel bandwidth, for example, when some other users of the broadband connection at the user premises 211 begin to use a large amount of bandwidth. If a newly generated image block/frame is also lost due to the discarded packet (even if FEC is used), in one embodiment, when the colocation service 210 is notified by the client 415: a second newly encoded image block The frame is discarded and the video compressor 404 reduces the data rate as it encodes a subsequently newly encoded image block/frame. Different embodiments use different techniques to reduce the data rate. For example, in one embodiment, this data rate reduction is achieved by reducing the quality of the encoded image block/frame, which is achieved by increasing the compression ratio. In another embodiment, the data rate is reduced by reducing the frame rate of the video (e.g., from 60 fps to 30 fps) and correspondingly slowing the rate of data transmission. In an embodiment, two techniques for reducing the data rate are used (eg, reducing the frame rate and increasing the compression ratio). If the lower data transmission rate successfully reduces the dropped packets, the colocation service 210 will continue to encode at a lower data rate according to the previously described channel data rate detection and adjustment method, and then allowed on the channel. The data rate is gradually adjusted up or down. Continuous reception of feedback data associated with discarded packets and/or delays allows the colocation service 210 to dynamically adjust the data rate based on current channel conditions.

State management in online gaming systems

One embodiment of the present invention uses techniques for efficiently storing the current state of an active game and porting the current state between servers. Although the embodiments described herein are related to online gaming, the basic principles of the present invention can be applied to various other types of applications (eg, designing applications, word processors, communications such as email or instant messaging). Software, etc.). Figure 27a illustrates an exemplary system architecture for implementing this embodiment, and Figure 27b illustrates an exemplary method. Although the method and the system architecture will be described simultaneously, the method illustrated in Figure 27b is not limited to any particular system architecture.

At 2751 of Figure 27b, a user initiates a new online game on the colocation service 210a from the client device 205. In response, at 2752, one of the game "game" images 2702a is self-contained (eg, a hard drive, directly connected to a server executing the game, or connected via a network to A server) is loaded into a memory (eg, RAM) on the colocation service 210a. The "game original" image contains the execution time code and data of the game prior to the initial play of any game (eg, like the first execution of the game). The user then plays the game at 2753 to change the "game" image to a non-game original image (e.g., an ongoing game represented by "State A" in Figure 27a). At 2754, the game is suspended or terminated by the user or colocation service 210a. At 2755, the state management logic 2700a on the colocation service 210a determines the difference between the game's "game" image and the current game state ("state A"). The difference between two binary images can be calculated using a variety of known techniques, including, for example, the techniques used in the well-known "diff" utility in Unix operating systems. Of course, the basic principles of the invention are not limited to any particular technique for differential calculation.

Regardless of how the differences are calculated, once the differences are calculated, the difference data is stored locally in the storage device 2705a and/or transmitted to a different colocation service 210b. If transmitted to a different colocation service 210b, the difference data may be stored on a storage device (not shown) at the new colocation service 210b. In any case, the difference data is associated with the user's account on the colocation service, so the next time the user logs into the colocation service and starts the game, the difference data can be identified. In an embodiment, instead of transmitting the difference data immediately, the difference data is transmitted to the next time the user attempts to play the game (and identifies a different colocation service as the best choice for the escrow game). A new hosting service.

Returning to the method illustrated in Figure 27b, at 2757, the user restarts the game from a client device, which may be the same client device 205 or a different client used by the user to initially play the game. Device (not shown). In response, at 2758, the state management logic 2700b on the colocation service 210b retrieves the "game" image of the game and the difference data from a storage device. At 2759, state management logic 2700b combines the game original image with the difference data to reconstruct the state of the game on the original colocation service 210a ("state A"). The difference data can be used to reconstruct the state of the binary image using various known techniques, including, for example, the techniques used in the well-known "patch" utility in Unix operating systems. You can also use the difference calculation technique used in familiar backup programs such as PC backup programs. The basic principles of the invention are not limited to any particular technique for reconstructing binary images using differential data.

Additionally, at 2760, platform dependency data 2710 is incorporated into the final game image 2701b. Platform dependency data 2710 may include any material that is only available on the destination server platform. By way of example and not limitation, the platform dependency data 2710 may include a media access control (MAC) address ^{, a} TCP/IP address, a time point, and a hardware serial number of the new platform (eg, a hardware serial number of the hard disk drive and the CPU). ), network server address (for example, DHCP/Wins server) and software serial number/boot code (including operating system serial number/startup code).

Other platform dependencies related to the client/user may include, but are not limited to, the following:

1. User's screen resolution. When the user restarts the game, the user may be using a different device with a different resolution.

2. User's controller configuration. When the game restarts, the user may have switched from a game controller to a keyboard/mouse.

3. User rights, such as whether a discount rate has expired (eg, the user has played a game during a promotion period, and now plays the game during a normal period at a higher cost), or whether the user or device is Have a certain age limit (for example, the parent of the user may have changed the settings for the child so that the child is not allowed to view the adult material, or the device playing the game (for example, a computer in a public library) has specific restrictions on whether adult materials can be displayed. ).

4. The level of the user. The user may have been allowed to play a multiplayer game in a particular league, but since some other users have exceeded the user's level, the user may have been downgraded to a lower level.

The foregoing examples of platform dependency data 2710 are provided to illustrate this embodiment of the invention. The basic principles of the invention are not limited to any particular setting of platform dependency data.

Figure 28 graphically illustrates how the state management logic 2700a of the first colocation service provides extraction of the difference profile 2800 from the ongoing game 2701a. The state management logic 2700b of the second colocation service then combines the game original image 2702b with the difference data 2800 and the platform dependency data 2710 to regenerate the state of the in-game 6701b. As generally shown in FIG. 28, the size of the difference data is significantly smaller than the size of the entire game image 2701a, and thus, a large amount of storage space and bandwidth are saved due to storage/transmission of only the difference data. Although not shown in FIG. 28, when the platform dependency data 2710 is incorporated into the last game image 2701b, the platform dependency data may overwrite some of the difference data.

Although an online video game implementation has been described above, the basic principles of the present invention are not limited to video games. For example, the aforementioned state management techniques can be implemented in the context of any type of online hosting application.

Technique for maintaining a client decoder

In one embodiment of the invention, whenever the user requests to connect to the colocation service 210, the colocation service 210 transmits a new decoder to the client device 205. Thus, in this embodiment, the decoder used by the client device is always the latest and uniquely suitable for hardware/software implemented on the client device.

As illustrated in Figure 29, in this embodiment, an application permanently installed on the client device 205 does not include a decoder. More specifically, the application is a client downloader application 2903 that manages the download and installation of the temporary decoder 2900 each time the client device 205 is connected to the colocation service 210. The downloader application 2903 can be implemented in hardware, software, firmware, or any combination thereof. In response to a user request for a new online session, the downloader application 2903 transmits information related to the client device 205 via a network (eg, the Internet). The information may include identification data identifying the hardware/software configuration (eg, processor, operating system, etc.) of the client device and/or the client device.

Based on this information, the downloader application 2901 on the hosting service 210 selects the appropriate temporary decoder 2900 to be used on the client device 205. The downloader application 2901 on the colocation service then transmits the temporary decoder 2900, and the downloader application 2903 on the client device verifies the decoder and/or installs the decoder on the client device 205. Encoder 2902 then encodes the audio/video content using any of the techniques described herein and transmits content 2910 to decoder 2900. Once the new decoder 2900 is installed, the decoder decodes the content of the current online session (i.e., using one or more of the audio/video decompression techniques described herein). In an embodiment, decoder 2900 is removed (eg, unmounted) from client device 205 when the session is terminated.

In one embodiment, when the temporary decoder 2900 is downloading, the downloader application 2903, by performing a data rate achievable on the channel (eg, by determining the time taken to download the data), the packet loss rate on the channel. The channel is characterized by a channel rating of the delay of the channel. The downloader application 2903 generates channel characterization data describing the channel ratings. This channel characterization data is then transmitted from the client device 205 to the colocation service downloader 2901, which uses the channel characterization data to determine how best to utilize the channel to transmit the media to the client device 205.

The client device 205 will typically send a message back to the colocation service 205 during the download of the temporary decoder 2900. Such messages may include a response message indicating that the packet was received without error or erroneously. In addition, the messages provide feedback to the downloader 2901 regarding the data rate (based on the rate at which the packet is received), the packet error rate (based on the percentage of packets that have received a received error), and the round trip delay of the channel. (Based on the amount of time used before the downloader 2901 receives feedback regarding a given packet that has been transmitted).

By way of example, if the data rate is determined to be 2 Mbps, the downloader can select a video window for the encoder 2902 that is smaller than the resolution of the data rate determined to be 5 Mbps (for example, 1280×720 at 60 fps). Resolution (for example, 640 x 480 at 60 fps). Depending on the packet loss rate, different Forward Error Correction (FEC) or packet structures can be selected.

If the packet loss is extremely low, the compressed audio and video can be transmitted without any error correction. If the packet loss is medium, the compressed audio and video can be transmitted using error correction coding techniques such as those described previously and illustrated in Figures 11a, 11b, 11c, and 11d. If the packet loss is very high, it can be determined that an audiovisual stream of appropriate quality cannot be transmitted, and the client device 205 can notify the user that the colocation service cannot be used via the communication channel (ie, "link"), or The client device may attempt to establish a different route to the colocation service with lower packet loss (as described below).

If the delay is low, the compressed audio and video can be transmitted at a low delay and a session can be established. If the delay is too high (eg, greater than 80 milliseconds), then for games that require low latency, the client device 205 can notify the user that the colocation service is not available via the link; the link is available, but the response to the user input The time will be slow or "lag"; alternatively, the user may attempt to establish a different route to the colocation service with a lower latency (as described below).

The client device 205 may attempt to connect to the colocation service 210 via another route via a network (eg, the Internet) to see if the damage is reduced (eg, whether packet loss is low, latency is low, or even data) Is the rate higher?) For example, the colocation service 210 can be geographically connected to the Internet from multiple locations (eg, the Los Angeles escrow center and the Denver escrow center), and due to the congestion of Los Angeles, it is possible There is a high packet loss, but Denver does not have a congestion. In addition, the colocation service 210 can connect to the Internet via a plurality of internet service providers (eg, AT&T and Comcast).

Due to congestion or other problems between the client device 205 and one of the service providers (e.g., AT&T), packet loss and/or high latency and/or constrained data rates may result. However, if the client device 205 is connected to the colocation service 210 via another service provider (eg, Comcast), the client device may be able to connect without congestion problems and/or have lower packet loss and/or have Lower latency and / or higher data rate. Therefore, if the client device 205 experiences a packet loss above a specified threshold (eg, a specified number of dropped packets in a specified duration) when downloading the temporary decoder 2900, a delay above a specified threshold is exceeded. And/or a data rate below a prescribed threshold, in an embodiment, the client device attempts to reconnect to the colocation service 210 via an alternate route (typically by connecting to a different IP address or Different domain names are used to determine if a preferred connection is available.

If the connection still experiences unacceptable damage after exhausting the alternate connection option, the situation may be that the local connection to the client device 205 of the Internet is being compromised, or the client device is too far from the colocation service 210. Far from reaching an appropriate delay. In this case, the client device 205 can notify the user that the colocation service is not available via the link; or the colocation service can be used but is damaging; and/or only certain types of low latency games/applications can be used .

Since this evaluation and potential improvement of the link characteristics between the colocation service 210 and the client device 205 occurs while downloading the temporary decoder, its reduction of the client device 205 must be spent separately on downloading the temporary decoder 2900 and evaluating The amount of time on the link characteristics. Nonetheless, in another embodiment, the evaluation and potential improvement of link characteristics is performed by the client device 205, separate from the download temporary decoder 2900 (e.g., by using dummy test data rather than decoder code). There are many reasons why this implementation may be a better implementation. For example, in some embodiments, the client device 205 is implemented partially or completely in hardware. Therefore, for these embodiments, it is not necessary to download the software decoder in nature.

Use standard-based image block size compression

As noted above, when image block based compression is used, the underlying principles of the invention are not limited to any particular image block size, shape or orientation. For example, in a DCT-based compression system (such as MPEG-2 and MPEG-4), the image block can be a macroblock (usually representing a block of 16 pixels by 16 pixels used in video compression). The size of the). This embodiment provides a very fine level of granularity for a pair of image blocks to function.

In addition, various types of image block patterns can be used regardless of the image block size. For example, Figure 30 illustrates an embodiment in which multiple I-image blocks are used in each R frame 3001-3004. A rotation pattern is used in which I image blocks are scattered throughout each R frame such that each four R frames produces a complete I frame. Dispersing the I image block in this way can reduce the impact of packet loss (limited loss to a small area of display).

The image blocks may also have a size that is suitable for the complete native structure of the basic compression algorithm. For example, if an H.264 compression algorithm is used, in one embodiment, the image block is set to the size of an H.264 "slice". This allows the techniques described herein to easily integrate the content of various different standard compression algorithms such as H.264 and MPEG-4. Once the image block size is set to a native compression structure, the same techniques as described above can be implemented.

Technology for streaming reversal and playback operations

As previously described in connection with FIG. 15, the uncompressed video/audio stream 1529 generated by the application/game server 1521-1525 can be compressed by the shared hardware compression 1530 simultaneously with multiple resolutions, resulting in multiple compressions. Video/audio streaming 1539. For example, one of the video/audio streams generated by the application/game server 1521 can be compressed by the shared hardware compression 1530 at 1280×720×60 fps and transmitted as the outgoing Internet traffic 1599 via the outgoing route 1540. To one user. The same video/audio stream can be simultaneously scaled down to a thumbnail size (eg, 200×113) by shared hardware compression 1530 via path 1552 to application/game server 1522 (or via delay buffer 1515) for display as A thumbnail 1600 of one of the thumbnail sets in FIG. When the thumbnail 1600 is scaled to the size 1800 (1280×720×60 fps) in FIG. 18 by the medium size 1700 in FIG. 17, the thumbnail stream is not decompressed, and the application/game server 1522 can decompress the positive A copy of a 1280 x 720 x 60 fps stream sent to the user of the application/game server 1521, and scaling the higher resolution video as if scaling the self-reduced size to a size of 1280 x 720 . This method has the advantage of reusing the compressed stream of 1280 x 720 twice. However, this method has several disadvantages: (a) if the amount of data transmitted by the user's Internet connection changes, the compressed video stream sent to the user may change in image quality, thereby generating an application/game. The image quality of the change seen by the "bystander" user of the server 1522, even if the user's internet connection does not change; (b) the application/game server 1522 must use the processing resources to decompress the entire 1280 x 720 image. And then scale the image (and possibly a resampling filter) to display a much smaller size (eg, 640×360) during scaling; (c) if due to limited internet connection bandwidth and / or the lost/damaged packet discards the frame, and the bystander "reverses" and "pauses" the video recorded in the delay buffer 1515, then the bystander will find that the discarded frame is missing in the delay buffer (This is especially true for the user's "stepping" frame by frame); and (d) if the bystander reverses to find a particular frame in the video recorded in the delay buffer, then the application/game is waiting The server 1522 must look for an I frame or I image block in front of the searched frame in the video stream recorded in the delay buffer, and then decompress all the P frames/image blocks until the desired Until the frame. This same restriction applies not only to the user of the "bystander" video/audio stream, but also to the user who views the video/audio stream (for example, "show off the clip") (including the generation of video/audio strings) User of the stream).

An alternate embodiment of the present invention solves these problems by compressing the video stream with more than one size and/or structure. As described herein, the best compression of a stream based on characteristics of the network connection (eg, data bandwidth, packet reliability) and the user's native client capabilities (eg, decompression capability, display resolution) ("live" streaming) to stream to the end user. Compressing other streams (referred to herein as "HQ" streams) with high quality, one or more resolutions and in a structure capable of playing video, and delivering the HQ streams and storing them on the server Center 210. For example, in one embodiment, the compressed HQ streams are stored on the RAID array 1515 and are used to provide functions such as pause, reverse, and other playback functions (eg, "bow-off clips", which may Spread to other users for viewing).

As illustrated in Figure 31a, an embodiment of the present invention includes an encoder 3100 capable of compressing a video stream in at least two formats: a format 3110 that periodically includes an I picture block or an I picture frame; and a I image block or I frame format 3111, unless it is due to interruption of the stream or because I block or I frame is determined to be smaller than I block or I frame (as described above) it is necessary to include I image Block or I frame. For example, the P-frame can be used to compress the "live" stream 3111 that is transmitted to the user when playing a video game (unless the I-picture block or I-frame is required or smaller, as described above). In addition, the encoder 3100 of this embodiment simultaneously compresses the live video stream 3111 in a second format. In an embodiment, the second format periodically includes an I picture block or an I frame (or a similar type of image format). ).

Although the above embodiment uses an I picture block, an I picture frame, a P picture block, and a P picture frame, the basic principles of the present invention are not limited to any particular compression algorithm. For example, any type of image format (where the frame depends on the previous or subsequent frame) can be used in place of the P image block or P frame. Similarly, any type of image format that is not dependent on the previous or subsequent frame may be substituted for the I block or I frame described above.

As noted above, the HQ stream 3110 includes a periodic I frame (e.g., in an embodiment, around every 12 frames). This situation is important because if the user would like to quickly reverse the stored video stream to a specific point, an I-picture block or I-frame is required. For a compressed stream with only P-frames (ie, the first frame of the sequence is not an I-frame), the decoder must return to the first frame of the sequence (this can take several hours) And decompress the P frame until the point the user wants to reverse. In the case that every 12 frames stored in the HQ stream 3110 have an I frame, the user may decide to reverse to a specific point, and the nearest previous I frame of the HQ stream does not precede the desired frame. More than 12 frames. Even if the maximum decoding rate of the decoder is instantaneous (for example, one-sixth of a second for a stream of 60 frames per second), it is far from an I frame 12 (frame) / 60 (frame / second) =1/5 seconds. Moreover, in many cases, the decoder can operate faster than instant, so, for example, at 2 times the instantaneous speed, the decoder can decode 12 frames in 6 frames, which is only "reverse" delay ten. One second. Needless to say, if the most recent previous I frame has many frames before a reversal point (for example, "reverse" will take one tenth of an hour (6 minutes)), even a fast decoder (for example, 10) More than instant) will also have an unacceptable delay. In another embodiment, a periodic I image block is used, and in this case, when the user tries to reverse, the decoder will look for the most recent previous image block before the reverse point and then start decoding from that point. The image block is decoded until all image blocks of the inversion point are decoded. Although the periodic I block or I frame results in less efficient compression than the complete elimination of the I frame, the colocation service 210 typically has a local available bandwidth and storage capacity that is more than sufficient for managing the HQ stream.

In another embodiment, the encoder 3100 encodes a periodic I image block or I frame having a P image block or a P frame (as previously described) but also previously a B image block or a B frame. HQ stream. As previously described, the B-frame is the frame before the I-frame and is based on the frame difference that is reversed from the I-frame in time. The B image block is the image block counterpart before an I image block and is based on the frame difference of the backing from the I image block. In this embodiment, if the desired inversion point is a B frame (or contains a B image block), the decoder will look for the nearest subsequent I frame or I image block and rewind in time to decode until the decoding is desired. Inverting the point, and then as the video playback progresses from the reversal point, the decoder will decode the B-frame, I-frame, and P-frame (or its image block counterpart) in the advancing continuous frame. The advantage of using a B-frame or a B-image block in addition to the Type I and Type P is that a higher quality at a given compression ratio is often achieved.

In yet another embodiment, the encoder 3100 encodes the HQ stream all into an I frame. One of the advantages of this method is that each inverting point is an I frame, and as a result, it is not necessary to decode other frames to reach the inverting point. One disadvantage is that the compressed data rate will be very high compared to I, P or I, P, B stream encoding.

Other video streaming playback actions (eg, fast or slow reverse, fast forward or slow forward, etc.) typically use periodic I-frames or I-image blocks (alone or in combination with P and/or B counterparts) It is easy to implement in practice, because in each case, the stream is played in a different order than the frame that advances frame by frame in time, and as a result, the decoder must find and decode one of the specific frames in the sequence (often arbitrary) of). For example, in the case of a very fast forward fast forward (eg, 100x speed), each subsequent frame displayed is 100 frames after the previous frame. Even for a decoder that executes at 10 times the instantaneous speed and decodes 10 frames in 1 frame time, 10x is still too slow to achieve a 100x forward fast forward. However, by using the periodic I frame or I block as described above, the decoder can search for the nearest available I frame or I block of the frame to be displayed next and decode only the intervening frame or image block to the target image. The point of the box.

In another embodiment, the I frame is encoded in the HQ stream with a consistent periodicity (eg, always every 8 frames), and the user can use the fast forward and reverse forward ratios of the I frame rate. The speed multiplier is the exact multiple of the periodicity of the I frame. For example, if the I frame periodically has 8 frames, the user can use the forward fast forward or reverse speeds to be 1X, 2X, 3X, 4X, 8X, 16X, 64X, and 128X and 256X. For a faster rate than the I frame, the decoder will first jump forward at this speed to the nearest I frame (which is the number of frames in advance) (for example, if the currently displayed frame is in an I The first 3 frames of the frame, at a speed of 128X, the decoder will jump to a frame with 128+3 frames in advance), and for each subsequent frame, the decoder will accurately fall every time. The selected speed jumps on an I frame to an exact number of frames (eg, at a selected speed of 128X, the decoder will jump 128 frames). Therefore, if all the speeds that are faster than the I-frame periodicity are the exact multiple of the periodicity of the I-frame, the decoder never has to decode any previous or subsequent frames to search for the desired frame, and it only has to decode each displayed picture. An I frame of the box. For speeds that are slower than I frames periodically (eg, 1X, 2X, 3X, 4X), or for multiples of faster non-I frame periodicity, for each frame displayed, the decoder searches Which frames require a minimum of additional new decoded frames to display the desired frame (which is one of the undecoded I-frames or an already decoded frame that can still be decoded (in RAM or other fast memory) )), and then decode the intervention frame as needed until the desired frame is decoded and displayed. For example, in the case of 4X forward fast forward, in a sequence of I and P codes with 8X I frame periodicity, if the current frame is a P map of a frame after an I frame In the box, the desired frame to be displayed is after 4 frames, and the to-be-displayed frame may be the fifth P frame after the previous I frame. If the currently displayed frame (which has just been decoded) is used as a point together, the decoder will have to decode more than 4 P-frames to display the desired frame. If the previous I frame is used, the decoder will have to decode 6 frames (the I frame and the next 5 P frames) to display the desired frame. (Obviously, in this case, it is advantageous to use the currently displayed frame to minimize the extra frames to be decoded.) Next, the next frame to be decoded (four frames in advance) can be The first P frame after an I frame. In this case, if the currently decoded frame is used as the starting point, the decoder will have to decode more than 4 frames (2 P frames, 1 I frame and 1 P frame). However, if the next I frame is used instead, the decoder only has to decode the I frame and the subsequent P frame. (Obviously, in this case, it is advantageous to use the next I frame as a starting point to minimize the extra frames to be decoded.) Therefore, in this example, the decoder can use the currently decoded frame as The starting point alternates with the use of a subsequent I frame as the starting point. As a general rule, regardless of the HQ video streaming mode (fast forward, reverse or step) and speed, the decoder can start from any frame, which can be an I frame or a previously decoded A frame that requires a minimum number of newly decoded frames to display the desired frame for each subsequent frame displayed in the playback mode and speed.

As illustrated in FIG. 31b, one embodiment of the colocation service 210 includes stream replay logic 3112 for managing user requests for replaying the HQ stream 3110. The stream replay logic 3112 receives a client request containing a video play command (eg, pause, reverse, from a specified point play, etc.), interprets the commands, and decodes the HQ stream 3110 from the specified point (appropriately from one The I frame begins with a previously decoded frame and then advances or rewinds to the specified point). In an embodiment, a decoded HQ stream is provided to an encoder 3100 (possibly from the same encoder 3100 if it can encode more than one stream at a time, or a separate encoder 3100), so The technique described in the method) recompresses the HQ stream and transmits it to the client device 205. The decoder 3102 on the client device then decodes and reproduces the stream, as described above.

In an embodiment, the stream replay logic 3112 does not decode the HQ stream and then causes the encoder 3100 to re-encode the stream. More specifically, the stream replay logic only streams the HQ stream 3110 directly from the specified point to the client device 205. The decoder 3102 on the client device 205 then decodes the HQ stream. Because the playback functions described herein typically do not have the same low latency requirements as playing an instant video game (eg, if the player only watches the previous gameplay, rather than actively playing the game), the higher quality HQ is usually Additional delays inherent in streaming can result in an unacceptable end-user experience (eg, video with higher latency but higher quality).

By way of example and not limitation, if the user is playing a video game, the encoder 3100 will provide a live stream of substantially all P frames, which is optimized for the user's connection and the local user end. (For example, about 1.4 Mbps at 640 x 360 resolution). At the same time, the encoder 3100 also compresses the video stream into one HQ stream 3110 in the colocation service 310, and stores the HQ stream in a local digital video decoding at 10 Mbps for, for example, 1280×720. On the RAID array (multiple I frames per 12 frames). If the user presses a "pause" button, the game will pause on the last decoded frame of the client and the screen will freeze. Then, if the user presses a "reverse" button, the stream replay logic 3112 will read the HQ stream 3110 from the DVR RAID starting from the most recent I frame or the available decoded frame, as described above. The stream replay logic 3112 will decompress the intervening P-frames or B-frames as needed, reorder the frames as needed, causing the play sequence to rewind at the desired reversal speed, and then (using prior techniques well known in the art) The image scaling technique resets the size of the frame to be decoded to be 640×360 from 1280×720, and the live stream encoder 3100 recompresses the reordering at a resolution of 640×360. Stream and transfer it to the user. If the user pauses again and then steps through the video to watch a sequence carefully, the HQ stream 3110 on the DVR RAID can use each frame for single stepping (even if the original live string) The stream may drop the frame due to any of the many reasons described herein. In addition, the quality of video playback will be quite high at each point in the HQ stream, and there may be points in the live stream that, for example, the bandwidth is corrupted, resulting in a temporary drop in compressed image quality. Although the user can accept a damaged image quality for a short period of time or moving images, if the user stops at a specific frame (or steps slowly and carefully) and carefully studies the frame, it may not be acceptable for damage. quality. By specifying a point in the HQ stream (eg, 2 minutes ago), the user also has the ability to fast forward and jump to a specific point. In the case of a live video stream with only P-frames or little (or unpredictable) I-frames, all such operations are not practical in terms of their complete generality and high quality.

In one embodiment, the user has a video window (not shown), such as Apple QuickTime with a "scrubber" (ie, left and right slider control), as early as the HQ stream has stored the video. Or the Adobe Flash video window, which allows the user to swipe forward and backward in the video stream. Although it appears to the user that it is "dragging" the live stream, in reality, the user is dragging the stored HQ stream 3110, then resetting the size of the HQ stream and recompressing it into a Live streaming. In addition, as previously mentioned, if the HQ stream is viewed by any other person at the same time, or viewed by the user at a different time, the resolution of the HQ stream can be encoded at a higher rate than the live stream (or The resolution of the HQ stream is viewed at a low level, and the quality will be as high as the quality of the live stream of the viewer, possibly achieving the quality of the HQ stream.

Thus, by simultaneously encoding the live stream (as described herein, in a suitable manner for its low latency, bandwidth and packet fault tolerance requirements) and an HQ stream with high quality, streaming playback action requirements, The user therefore has the configuration to be configured in two situations. And, in fact, two different streams are encoded differently, which is effectively transparent to the user. From the user's point of view, although performing on a highly variable and relatively low bandwidth internet connection, it experiences extremely fast response with low latency, while digital video recording (DVR) functionality is flexible and flexible. The speed is of very high quality.

As a result of the above techniques, the user gets the benefit of both live streaming and HQ streaming during online gaming or other online interactions without suffering either of live streaming or HQ streaming restrictions.

Figure 31c illustrates an embodiment of a system architecture for performing the above operations. As illustrated, in this embodiment, encoder 3100 encodes a series of "live" streams 3121L, 3122L, and 3125L and a series of corresponding "HQ" streams 3121H1-H3, 3122H1-H3, and 3125H1-H3, respectively. Each HQ stream H1 is encoded at full resolution, and each encoder H2 and H3 scales the video stream to a smaller size prior to encoding. For example, if the resolution of the video stream is 1280×720, H1 is encoded at a resolution of 1280×720, and H2 can be scaled to 640×360 and encoded at the resolution, and H3 can be scaled to 320×180 and encoded with this resolution. Any number of simultaneous Hn scalers/encoders can be used to provide multiple simultaneous HQ encodings in multiple resolutions.

As described above, each of the live streams operates in response to channel feedback signals 3161, 3162, and 3165 received via the incoming Internet connection 3101 (see, for example, feedback in Figures 25-26). Discussion of signals 2501 and 2601). The live streams are transmitted via the Internet (or other network) via the outbound routing logic 3140. The live compressors 3121L-3125L include logic for adapting the compressed video streams (including scaling, dropping frames, etc.) based on channel feedback.

The HQ streams are routed by the incoming routing logic 3141 and 1502 via signal path 3151 to an internal delay buffer (eg, RAID array 3115) or other data storage device and/or fed back to the application/game servo via signal path 3152. And encoder 3100 for additional processing. As described above, the HQ stream 3121H n -3125H n is then streamed to the end user after the request (see, for example, Figure 31b and related text).

In one embodiment, encoder 3100 is implemented with shared hardware compression logic 1530 as shown in FIG. In another embodiment, some or all of the encoders and scalers are individual subsystems. The basic principles of the invention are not limited to scaling or compressing any particular sharing of resources or hardware/software configurations.

One advantage of the configuration of Figure 31c is that the application/game server 3121-3125, which requires less than a full size video window, will not have to process and decompress a full size window. In addition, the intermediate window sized application/game server 3121-3125 can receive a compressed stream that is close to the desired window size and then scale up or down to the desired window size. In addition, if multiple application/game servers 3121-3125 request the same size video stream from another application/game server 3121-3125, the incoming route 3141 can implement an IP multicast technology (such as this item). IP multicast technology well known in the art) and simultaneously broadcast the requested stream to multiple application/game servers 3121-3125 without the need to broadcast separate streams to each application/game server making the request . If a broadcast application/game server changes the size of a video window, it can be converted to broadcast a different video size. Therefore, any large number of users can simultaneously watch an application/game server video stream. Each user has the flexibility to scale the video window proportionally and always obtain a video stream that is strictly scaled to the desired window size. The benefits.

One disadvantage of the method shown in Figure 31c is that in many practical implementations of the colocation service 210, absolutely no simultaneous viewing of all compressed HQ streams (not to mention the full size of all compressed HQ streams) time. This waste is reduced when the encoder 3100 is implemented as a shared resource (e.g., a scaler/compressor, implemented in software or hardware). However, due to the bandwidth involved, there may be a practical problem of connecting a large number of uncompressed streams to a common shared resource. For example, each 1080p60 stream is almost 3 Gbps, which is far more than a billion bit Ethernet. The following alternative embodiment addresses this issue.

Figure 31d shows an alternate embodiment of the colocation service 210, wherein each application/game server 3121-3125 has two compressors assigned to it: (1) a live stream compressor 3121L-3125L, The compressed video stream is adapted based on channel feedback 3161-3165, and (2) an HQ stream compressor that outputs a full resolution HQ stream, as described above. It is worth noting that the live stream compressor is dynamic and adaptive, thereby utilizing two-way communication with the client 205, while the HQ stream is non-adaptive and unidirectional. The other difference between streams is that the quality of the live stream varies significantly depending on the channel conditions and the nature of the video material. Some frames may be of poor quality and may have frames discarded. In addition, the live stream can be almost all P frames or P blocks, while I frames or I blocks are rare. HQ streams typically have a much higher data rate than live streams, and HQ streams will provide consistent high quality without discarding any frames. The HQ stream can be all I-frames, or can have frequent and/or regular I-frames or I-picture blocks. The HQ stream can also include a B frame or a B image block.

In one embodiment, the shared video is scaled and recompressed 3142 (described in more detail below) to select only a particular HQ video stream 3121H1-3125H1 before being sent to the incoming route 3141 for routing as previously described. Scaled and recompressed with one or more different resolutions. The other HQ video stream passes through its full size to reach the incoming route 3141 for routing as previously described, or does not pass completely. In one embodiment, the pair is determined based on whether there is an application/game server 3121-3125 requesting a particular HQ stream at a particular resolution (or a resolution close to the scaled resolution or full resolution). The decision to scale and recompress which HQ streams and/or which HQ streams pass completely. By this means, the HQ stream that is scaled and recompressed (or possibly fully passed) is only the actual required HQ stream. In many applications of the colocation service 210, this results in a drastic reduction in scaling and compression resources. In addition, if each HQ stream is compressed by the compressor 3121H1-3125H1 at least with its full resolution, the bandwidth ratio to be delivered to and within the shared video scaling and recompression 3142 is not accepted. The bandwidth required for compressed video is significantly reduced. For example, a 3 GBps uncompressed 1080p60 stream can be compressed to 10 Mbps and still maintains extremely high quality. Therefore, for one-billion-bit Ethernet connectivity, it is not impossible to carry an uncompressed 3 Gbps video stream, which may carry many 10 Mbps video streams without significant degradation in quality.

Figure 31f shows the details of the shared video scaling and recompression 3142 and a number of HQ video compressors HQ 3121H1-3131H1. Based on the request from the application/game server 3121-3125 to scale up to a particular video stream of a particular size, the internal route 3192 typically selects one of the compressed HQ streams from the HQ video compressor HQ 3121H1-3131H1. Subset. A stream in the selected subset of the stream is delivered via decompressor 3161-3164 if the requested stream is scaled, or has full resolution at the requested stream In the case, it is delivered on the unscaled video path 3196. The stream that is waiting to be scaled will be decompressed by the decompressors 3161-3164 into uncompressed video, and then each uncompressed video is scaled to the requested size by the scalers 3171-3174. Each scaled video is then compressed by compressors 3181-3184. Note that if a particular HQ stream is requested at more than one resolution, then internal route 3192 (using the IP multicast technology well known to those skilled in the art) multicasts the stream to one or more solutions. The compressors 3161-3164 and (in the case where the stream has a requested size at full resolution) are multicast to the outgoing route 3193. All such requested streams (whether scaled (from compressors 3181-3184) or unscaled (from internal route 3192) are then sent to the outbound route 3193. Route 3193 then sends each requested stream to the application/game server 3121-3125 that requested the stream. In one embodiment, if more than one application/game server requests the same stream at the same resolution, then outgoing route 3193 multicasts the stream to all application/game servers 3121 that issued the request. -3125.

In the presently preferred embodiment of the shared video scaling and recompression 3142, the delivery is performed using a one billion bit Ethernet switch, and decompression, scaling, and compression are performed by each function. Discrete specialized semiconductor devices to implement. The same functionality can be implemented in hardware with a high degree of integration or by an extremely fast processor.

31e shows another embodiment of a colocation service 210 in which the functionality of the delay buffer 3115 (described previously) is implemented in a shared video delay buffer, scaling and decompression subsystem 3143. The details of subsystem 3143 are shown in Figure 31g. Except for the following operations, the operation of subsystem 3143 is similar to the operation of subsystem 3142 shown in Figure 31f: 3191 first selects the HQ video stream to be delivered based on the request from the application/game server 3121-3125, and The request-delayed HQ stream is then delivered via a delay buffer 3194 implemented in a current preferred embodiment as a RAID array (but can be implemented in any storage medium having sufficient bandwidth and capacity), and without delay The video path 3195 delivers a stream of unsolicited delays. The output of delay buffer 3194 and undelayed video 3195 is then routed by internal route 3192 based on whether the requested stream is scaled or not scaled. The scaled stream is routed to the outgoing route 3193 via the decompressors 3161-3164, the scalers 3171-3174, and the compressors 3181-3184, and the unscaled video 3196 is also sent to The outgoing route 3193, and then the outgoing route 3193, then sends the video to the application/game server in unicast or multicast mode in the same manner as previously described in subsystem 3142 of Figure 31f.

Another embodiment of a video delay buffer, scaling and decompression subsystem 3143 is shown in Figure 31h. In this embodiment, an additional delay buffer HQ 3121D-HQ 3131D is provided for each HQ stream. If the cost of RAM and flash ROM (which can be used to delay an otherwise compressed video stream) drops rapidly, this can become less expensive and/or more flexible than having a shared delay buffer 3194. Alternatively, in yet another embodiment, a single delay buffer 3197 (shown in dashed lines) can provide all HQ streams individually in a high performance collective resource (eg, very fast RAM, flash memory, or disk). Delay. In any scenario, each delay buffer HQ 3121D-3131D can variably delay a stream from the HQ video source or cause the stream to pass without delay. In another embodiment, each delay buffer can provide multiple streams with different amounts of delay. All delays or no delays are requested by the application/game services 3121-3125. In all such cases, the delayed and undelayed video stream 3198 is sent to internal route 3192 and proceeds through the remainder of subsystem 3143 as previously described with respect to Figure 31g.

In the previous embodiment with respect to various figures 31n, note that live streaming utilizes a two-way connection and is tailored for a particular user with minimal delay. HQ streaming utilizes one-way connections and is both unicast and multicast. Please note that although the multicast function is illustrated in this figure as a single unit (such as in a one-bit Ethernet switch), in a large system, the multicast function is likely to A tree implementation with multiple switches. In fact, in the case of a video stream from a top video game player, the situation is likely to be as follows: millions of users simultaneously watch the player's HQ stream. In this case, there may be many individual switches in the successor stage broadcasting the multicast HQ stream.

For diagnostic purposes and to provide feedback to the user (eg, to let the user know the popularity of his game show), in one embodiment, the colocation service 210 will record the video of each application/game server 3121-3125. How many viewers are there at the same time. This recording can be accomplished by maintaining an execution count of the number of valid requests for a particular video stream by the application/game server. Therefore, a player with 100,000 spectators at the same time will know that his gameplay is very popular, and it will be a motivation for gamers to better perform and attract viewers. When video streaming (e.g., a video game tournament) has a large number of television viewers, it may be desirable for the reviewer to speak during the video game competition so that some or all of the users viewing the multicast can hear their commentary.

Applications and games executed on such application/game servers should have an application interface (API) in which applications and/or games can be submitted with specific characteristics (eg, resolution and latency). A request for a specific video stream. In addition, such APIs that are subject to a job environment executing on an application/game server or a hosted service control system of Figure 4a may reject such requests for a variety of reasons. For example, the requested video stream may have specific permission restrictions (eg, such that it can only be viewed by a single viewer without being broadcast to other viewers), there may be subscription restrictions (eg, the viewer may have to be able to view The streaming payment) may have an age limit (for example, the viewer may have to be 18 years old to view the stream), and there may be privacy restrictions (for example, a person using the app or playing the game may limit viewing to only a selected number) Or a category of viewers (eg, "friends"), or may not be allowed to watch at all, and there may be restrictions on requesting delay material (eg, in the event that the user is playing a theft game that may reveal their location). There are many other limitations that may limit the viewing of streaming. In either of these cases, the application/game server request may be rejected in the event that the reason for the rejection is specified, and in an embodiment, the request is accepted in an alternate manner (eg, the description must be subscription) The fee paid).

The HQ video stream stored in the delay buffer in any of the previous embodiments may be exported to other destinations other than the colocation service 210. For example, a particularly interesting video stream can be requested by an application/game server (usually upon request by a user) to export to YouTube. In this case, the video stream can be transmitted over the Internet in a YouTube-approved format along with appropriate descriptive information (eg, the name, time, score, etc. of the user playing the game). This transmission can be implemented by multicasting the commentary audio to all Game/App servers 3121-3125 requesting this narration in a separate stream. The Game/App servers can incorporate the narrated audio into the audio stream sent to the user premises 211 using audio mixing techniques well known to those skilled in the art. There must be multiple commentators (eg, having different opinions, or using different languages), and the user can make a selection among them.

In a similar manner, individual audio streams can be mixed in or act as a substitute for a particular video stream (or individual stream) in the colocation service 210, thereby mixing or replacing the stream from the instant stream. Or audio from a delay buffer video. This audio can be an narration or narration, or the audio can provide a sound for the character in the video stream. This allows the mechanical animation (the animation generated by the user's video game video stream) to be easily generated by the user.

The video stream described in this document is shown as being captured from the video output of the application/game server and then streamed and/or delayed and reused or distributed in various ways. The same delay buffer can be used to save video material from non-application/game server sources and provide the same degree of playback and distribution flexibility (with appropriate limits). Such sources include live feeds from television stations (either in the air or in a non-air manner (such as CNN), and or charges (such as HBO) or free). Such sources also include pre-recorded movies or television shows, home movies, advertisements, and live video conference call feeds. The live feed can be handled like a live output of a Game/App server. The pre-recorded material can be disposed of like a delay buffer output.

In one embodiment, the various functional modules and associated steps described herein may be performed by a particular hardware component containing fixed-line logic (such as a special application integrated circuit (ASIC)) for performing the steps. Or by any combination of stylized computer components and custom hardware components.

In one embodiment, the modules can be implemented on a programmable digital signal processor ("DSP") such as Texas Instruments' TMS320x architecture (eg, TMS320C6000, TMS320C5000, etc.). A variety of different DSPs can be used while still complying with these basic principles.

Embodiments may include various steps as set forth above. The steps may be embodied as machine executable instructions that cause a general purpose or special purpose processor to perform the specific steps. Various components (such as computer memory, hard disk drives, input devices) that are not related to these basic principles have been omitted from some or all of the figures to avoid obscuring the relevant aspects.

The disclosed elements can also be provided as a machine readable medium for storing machine executable instructions. The machine-readable medium can include, but is not limited to, a flash memory, a compact disc, a CD-ROM, a DVD ROM, a RAM, an EPROM, an EEPROM, a magnetic or optical card, a broadcast medium, or other type of machine suitable for storing electronic instructions. Read the media. For example, the present invention can be downloaded as a computer program via a communication link (eg, a data modem or network connection) by means of a data signal embodied in a carrier or other broadcast medium. A remote computer (eg, a server) is transmitted to a requesting computer (eg, a client).

It should be understood that the disclosed elements can also be provided as a computer program product, which can include a machine readable medium storing instructions, which can be used to program a computer (eg, a processor or Other electronic devices) to perform a series of operations. Alternatively, the operations can be performed in combination with one of the hardware and the software. The machine-readable medium can include, but is not limited to, floppy disks, optical disks, CD-ROMs and magneto-optical disks, ROM, RAM, EPROM, EEPROM, magnetic or optical cards, media, or other types of media suitable for storing electronic instructions. Machine readable medium. For example, the disclosed elements can be downloaded as a computer program product, where the program can be embodied in a carrier or other communication via a communication link (eg, a data machine or network connection) The data signals in the media are transmitted from a remote computer or electronic device to a request handler.

In addition, while the disclosed subject matter has been described in connection with the specific embodiments, many modifications and variations are still within the scope of the invention. Therefore, the description and drawings are considered from the meaning of the description rather than the meaning of limitation.

100. . . CPU/GPU

101. . . Random access memory (RAM)

102. . . Display device

103. . . Hard disk drive

104. . . CD player

105. . . Network connection

106. . . Game Controller

205. . . Client device

206. . . Internet

210. . . Hosting service

210a. . . Hosting service

210b. . . Hosting service

211. . . User location

220. . . Game or app developer

221. . . External input device

222. . . Internal or external display device

301. . . Nominal maximum data rate

302. . . Actual maximum data rate available

303. . . Data rate

401. . . Hosted service control system

402. . . server

403. . . Storage Area Network (SAN)

404. . . Low latency video compression logic

405. . . Redundant Array of Independent Disks (RAID)

406. . . control signal

406a. . . control signal

406b. . . control signal

409. . . Routing logic

410. . . Internet

412. . . Low latency video decompression logic

413. . . Control signal logic

415. . . user terminal

421. . . Input device

422. . . Display device

440. . . Internet bypass service

441. . . WAN interface

442. . . WAN interface

443. . . Firewall/router/network address translation (NAT) device

451. . . control signal

452. . . User location routing

453. . . User ISP round trip delay

454. . . Internet round trip delay

455. . . One-way routing delay

456. . . Maximum one-way delay

457. . . One-way delay

458. . . Video decompression delay

462. . . Ethernet socket

463. . . HDMI connector

464. . . Display capability

465. . . Home or office client device

466. . . Shutter glasses

468. . . Display device / standard definition television (SDTV) or high definition television (HDTV) / monitor

469. . . Input device

472. . . Internet connection

475. . . Home or office client device

476. . . Flash memory

477. . . Video camera

478. . . Display device

479. . . Bluetooth Input Device / Wired USB Input Device

480. . . Busbar

481. . . Ethernet interface

482. . . WiFi subsystem

483. . . Control CPU

484. . . Bluetooth interface

485. . . USB subsystem

486. . . Video decompressor

487. . . Video output subsystem

488. . . Audio decompression subsystem

489. . . Audio output subsystem

490. . . High Definition Multimedia Interface (HDMI) interface

491. . . DVI-I

492. . . S video

493. . . Composite video

494. . . Digital interface

495. . . Stereo analog interface

497. . . Ethernet with Power over Ethernet (PoE)

499. . . power supply

501. . . Individual video frame

502. . . Individual video frame

503. . . Individual video frame

511. . . Compressed frame

512. . . Compressed frame

513. . . Compressed frame

520. . . Compression logic

560. . . Uncompressed video frame

561. . . Uncompressed video frame

611. . . I frame

612. . . P frame

613. . . P frame

620. . . Compression logic

621. . . Rated maximum data rate

622. . . Maximum data rate available

623. . . Peak data rate required for I frame

624. . . Video streaming data rate

633. . . I frame peak

634. . . Video stream data rate sequence

635. . . P frame peak

636. . . P frame peak

670. . . B frame

671. . . I frame

701-760. . . Uncompressed frame

711-770. . . R frame

721-770. . . Uncompressed frame

731-780. . . R frame

805. . . Character

811. . . R frame

812. . . R frame

922. . . Maximum data rate available

934. . . Video streaming data rate

941. . . Peak data rate

942. . . 2 times peak data rate

943. . . 3 times peak data rate

944. . . 4 times peak data rate

952. . . 2 times peak

953. . . Flatten 2 times peak

954. . . Flatten 4 times peak

955. . . Flatten 4 times peak

961-970. . . Uncompressed frame

981. . . Compressed frame 1

982. . . Compressed frame 2

983. . . Compressed frame 3

985. . . Compressed frame 5

986. . . Compressed frame 6

987. . . Compressed frame 10

991. . . Transmission time

992. . . Transmission time

993. . . Transmission time

995. . . Transmission time

996. . . Transmission time

997. . . Transmission time

1001. . . Packet

1002. . . Packet

1003. . . Packet

1004. . . Packet

1005. . . Packet

1010. . . Image block packaging logic

1100. . . I image block

1101. . . Forward Error Correction (FEC) Code

1103. . . I image block

1104. . . FEC code

1105. . . FEC code

1110. . . Audio packet

1111. . . FEC code

1112. . . Audio packet

1120. . . User input command

1121. . . FEC code

1122. . . User input command

1123. . . FEC code

1200. . . Multi-core and / or multi-processor

1201-1208. . . core

1205. . . 4x2 configuration

1300. . . Hosting service 210 server center

1301. . . network

1501. . . Internet traffic

1502. . . Inbound routing

1511. . . Redundant Array of Independent Disks (RAID)

1512. . . RAID array

1515. . . RAID array

1521. . . Application/game ("app/game") server

1522. . . Application/game server

1525. . . Application/game server

1529. . . Uncompressed video/audio

1530. . . Shared video compression / shared hardware compression

1539. . . Compressed video/audio

1540. . . Routing

1550. . . Delay buffer and/or group video video path

1551. . . path

1552. . . path

1599. . . Outgoing Internet traffic

1600. . . Thumbnail video window

1601. . . Yellow selection box

1602. . . Game information

1700. . . Video window

1800. . . arrow

1801. . . Overlay

1900. . . Video and audio

2000. . . Video

2001. . . Live video

2002. . . Video clip

2003. . . Video clip / show off clip

2004. . . Show a number showing how many times the clip has been viewed

2005. . . Keyhole shape icon

2400. . . line

2401. . . 1.5Mbps connection speed

2402. . . 12 Mbps connection speed

2403. . . 96 Mbps connection speed

2500. . . Encoder

2501. . . Feedback channel

2502. . . decoder

2510. . . Frame

2600. . . Encoder

2601. . . Feedback channel

2602. . . decoder

2605. . . Uncompressed image block/frame

2606. . . P image block / frame

2607. . . Decompressed image block/frame

2610. . . Memory device

2611. . . Past state of the encoder

2620. . . Memory device

2621. . . Past state of the decoder

2700a. . . State management logic

2700b. . . State management logic

2701a. . . Executive game

2701b. . . Final game image

2702a. . . Game "Original Game" image

2702b. . . Game original image

2705a. . . Storage device

2710. . . Platform dependency data

2800. . . Difference data

2900. . . decoder

2901. . . Downloader application

2902. . . Encoder

2903. . . Client downloader application

2910. . . content

3001. . . R frame

3002. . . R frame

3003. . . R frame

3004. . . R frame

3100. . . Encoder

3101. . . Internet access

3102. . . decoder

3110. . . HQ stream

3111. . . "live" streaming

3111. . . RAID array

3112. . . Stream replay logic

3112. . . RAID array

3115. . . RAID array

3121~3125. . . Application/game server

3121H. . . "HQ" stream

3121L. . . "live" streaming

3122H. . . "HQ" stream

3122L. . . "live" streaming

3125H. . . "HQ" stream

3125L. . . "live" streaming

3140. . . Routing logic

3141. . . Routing logic

3142. . . Shared video scaled and recompressed

3143. . . Shared video delay buffer, scaling and decompression subsystem

3151. . . Signal path

3152. . . Signal path

3161. . . Decompressor

3161~3165. . . Channel feedback signal

3162. . . Decompressor

3163. . . Decompressor

3164. . . Decompressor

3171. . . Proportional adjuster

3172. . . Proportional adjuster

3173. . . Proportional adjuster

3174. . . Proportional adjuster

3181. . . compressor

3182. . . compressor

3183. . . compressor

3184. . . compressor

3191. . . Inbound routing

3192. . . Internal routing

3193. . . Routing

3194. . . Delay buffer

3195. . . Undelayed video path

3196. . . Unscaled video path

3197. . . Single delay buffer

3198. . . Delayed and undelayed video streaming

HQ 3121D~HQ 3131D. . . Delay buffer

HQ 3121H1~HQ 3131H1 HQ. . . Video compressor

HS-1-HS-6. . . Server center

T1-T8. . . Image block

Figure 1 illustrates the architecture of a prior art video game system.

2a-2b illustrate a high level system architecture in accordance with an embodiment.

Figure 3 illustrates the actual, nominal and required data rate of communication between a client and a server.

Figure 4a illustrates the use of a colocation service and a client in accordance with an embodiment.

Figure 4b illustrates an exemplary delay associated with communication between a client and a colocation service.

Figure 4c illustrates a client device in accordance with an embodiment.

Figure 4d illustrates a client device in accordance with another embodiment.

Figure 4e illustrates an example block diagram of the client device of Figure 4c.

Figure 4f illustrates an example block diagram of the client device of Figure 4d.

Figure 5 illustrates an example form of video compression that may be used in accordance with an embodiment.

Figure 6a illustrates an example form of video compression that may be used in another embodiment.

Figure 6b illustrates the peak of the data rate associated with transmitting a low complexity, low activity video sequence.

Figure 6c illustrates the peak of the data rate associated with transmitting a high complexity, high activity video sequence.

Figures 7a through 7b illustrate an example video compression technique used in an embodiment.

Figure 8 illustrates an additional example video compression technique used in an embodiment.

Figures 9a through 9c illustrate a frame rate processing technique used in an embodiment of the present invention.

Figures 10a through 10b illustrate an embodiment in which an image image block is efficiently packaged within a package.

Figures 11a through 11d illustrate an embodiment using a forward error correction technique.

Figure 12 illustrates an embodiment in which multiple multi-core processing units are used for compression.

Figures 13a-13b illustrate geolocation and communication between colocation services, in accordance with various embodiments.

Figure 14 illustrates an exemplary delay associated with communication between a client and a colocation service.

Figure 15 illustrates an example colocation service server central architecture.

Figure 16 illustrates an example screen capture of an embodiment of a user interface including a plurality of live video windows.

Figure 17 illustrates a diagram of the user interface of Figure 16 after selecting a particular video window.

Figure 18 illustrates the user interface of Figure 17 after scaling the particular video window to full screen size.

Figure 19 illustrates an example collaborative user video material overlaid on a multiplayer game screen.

Figure 20 illustrates an example user page of a game player on a colocation service.

Figure 21 illustrates an example 3D interactive advertisement.

Figure 22 illustrates an example sequence of steps for producing a photorealistic image of a textured surface from a surface of a live performance.

Figure 23 illustrates an example user interface page that allows selection of linear media content.

Figure 24 is a graph showing the relationship between the amount of time elapsed before the live broadcast of the web page and the connection speed.

Figures 25a through 25b illustrate an embodiment of the feedback channel of the present invention using a client device to a colocation service.

Figures 26a-26b illustrate an embodiment of encoding an image block/frame based on the last known image block/frame that has been successfully received.

27a-27b illustrate an embodiment of porting a game or application state from a first colocation service or server to a second colocation service or server.

Figure 28 illustrates an embodiment of a state in which a game or application is migrated using the difference data.

Figure 29 illustrates an embodiment of the present invention using a temporary decoder on a client device.

Figure 30 illustrates how an "I-image block" is spread across an "R-frame" in accordance with an embodiment of the present invention.

31a through 31h illustrate an embodiment of the present invention that produces a live stream and/or one or more HQ streams.

404. . . Low latency video compression logic

1200. . . Multi-core and / or multi-processor

1201-1208. . . core

1205. . . 4x2 configuration

Claims (12)

A computer-implemented method for performing video compression, comprising: receiving user input associated with an online video game or application, the user input being transmitted by a user from a client over the Internet; And responsively executing the video game or application code to reproduce a sequence of video images caused by execution of the video game or application; encoding a plurality of video frames or portions thereof according to a first encoding format, The first encoding format is optimized for transmission to a client device via a current communication channel; the plurality of encoded video frames or portions are transmitted to the client device via the communication channel; The second encoding format encodes the first plurality of video frames, and the second encoding format has a relatively higher quality compressed video and/or a lower compression ratio than the first encoding format; The first plurality of video frames encoded in the second encoding format are stored on a storage device; and the first plurality of video frames encoded in the second encoding format are encoded The frame may be used by the client device and other client devices to play a video stream, wherein the first plurality of video frames are a video game generated when a user of the user device plays an online video game. a frame, and wherein the first encoding format is selected to reduce a delay in transmitting the frames compressed according to the first encoding format to the client device via the current communication channel, and wherein the second encoding format is selected Seal the line The video frames of the video game enable subsequent playback of the video frames from the archive. The method of claim 1, wherein the first encoding format comprises a frame that depends on other frames, and the second encoding format includes some frames that are not dependent on other frames. The method of claim 2, wherein the second encoding format comprises a plurality of I frames, and wherein the first encoding format does not include an I frame. The method of claim 3, wherein the first encoding format only includes a P-frame. A computer system having a memory for storing a code and a processor for processing the code to perform a process of receiving user input associated with an online video game or application, the user The input is transmitted by a user from a client over the Internet, and the code of the video game or application is responsively executed to reproduce a sequence of video images caused by the execution of the video game or application; A first encoding format encoding a plurality of video frames or portions thereof, the first encoding format being optimized for transmission to a client device via a current communication channel; the plurality of encoded devices being encoded via the communication channel Transmitting a video frame or portion to the client device; encoding the first plurality of video frames in parallel according to a second encoding format, the second encoding format having a relatively higher quality than the first encoding format Compressing video and/or a lower compression ratio; storing the first plurality of video frames encoded in the second encoding format Stored on a storage device; and the first plurality of video frames encoded in the second encoding format can be used by the user equipment and other client devices to play video streams, wherein the first plurality of video frames The frame is a frame of a video game generated when a user of the user device plays an online video game, and wherein the first encoding format is selected to reduce compression according to the first encoding format via the current communication channel transmission The delay of the frame to the user device, and wherein the second encoding format is selected to archive the video frames of the online video game, so that the video frames can be played back from the archived file. The computer system of claim 5, wherein the first encoding format comprises a frame that is dependent on the other frames, and the second encoding format includes some frames that are non-dependently dependent on the other frames. The computer system of claim 6, wherein the second encoding format comprises a plurality of I frames, and wherein the first encoding format does not include an I frame. The computer system of claim 7, wherein the first encoding format only includes a P-frame. A machine readable medium storing code that, when executed by a machine, causes the machine to: receive user input associated with an online video game or application, the user input being The user transmits the video game or the application code from a client through the Internet to reproduce the sequence of the video image caused by the execution of the video game or the application; Encoding a plurality of video frames or portions thereof according to a first encoding format, the first encoding format being optimized for transmission to a client device via a current communication channel; encoding the plurality of encodings via the communication channel The video frame or portion is transmitted to the user equipment; the first plurality of video frames are encoded in parallel according to a second encoding format, and the second encoding format has a relatively higher level than the first encoding format. a compressed video of quality and/or a lower compression ratio; storing the first plurality of video frames encoded in the second encoding format on a storage device; and causing the encoding in the second encoding format A plurality of video frames can be used by the user equipment and other client devices to play video streams, wherein the first plurality of video frames are generated when a user of the user device plays an online video game. a frame of a video game, and wherein the first encoding format is selected to reduce transmission of the frames compressed according to the first encoding format to the user device via the current communication channel The time delay, and wherein the second encoding format is selected to archive the video frames of the online video game, such that the video frames can be played back from the archived file. The machine readable medium of claim 9, wherein the first encoding format comprises a frame that is dependent on other frames, and the second encoding format comprises some frames that are non-dependently dependent on the other frames. The machine readable medium of claim 10, wherein the second encoding format comprises a plurality of I frames, and wherein the first encoding format does not include an I frame. The machine readable medium of claim 11, wherein the first encoding format comprises only P frames.